WHO operational handbook on tuberculosis: module 4: treatment: drug-susceptible tuberculosis treatment: web annex 2: optimization of dosage of the first-line medicines rifampicin, isoniazid, ethambutol and pyrazinamide in treatment of drug-susceptible tuberculosis: summary of evidence from four systematic reviews

WHO operational handbook on tuberculosis Web annex 2. Optimization of dosage of the first-line medicines rifampicin, isoniazid, ethambutol and pyrazinamide in treatment of drug-susceptible tuberculosis: summary of evidence from four systematic reviews Module 4: Treatment Drug-susceptible tuberculosis treatment

Optimization of dosage of the first-line medicines rifampicin, isoniazid, ethambutol and pyrazinamide in treatment of drug-susceptible tuberculosis: summary of evidence from four systematic reviews

Optimization of dosage of the first- line medicines rifampicin, isoniazid, ethambutol and pyrazinamide in treatment of drug-susceptible tuberculosis: summary of evidence from four systematic reviews SUMMARY REPORT K. Haigh1, H. Twabi2, H. Ryan1, V. Lutje3, S. Nevitt1 and G. Davies1 1 University of Liverpool 2 College of Medicine, University of Malawi 3 Liverpool School of Tropical Medicine WHO operational handbook on tuberculosis. Module 4: treatment - drug-susceptible tuberculosis treatment. Web Annex 2. Optimization of dosage of the first-line medicines rifampicin, isoniazid, ethambutol and pyrazinamide in treatment of drug-susceptible tuberculosis: summary of evidence from four systematic reviews/ Kathryn Haigh, Hussein Twabi, Hannah Ryan, Vittoria Lutje, Sarah Nevitt, Gerry Davies ISBN 978-92-4-005079-2 (electronic version) © World Health Organization 2022 Some rights reserved. This work is available under the Creative Commons Attribution-NonCommercial-ShareAlike 3.0 IGO licence (CC BY-NC-SA 3.0 IGO; https://creativecommons.org/licenses/by-nc-sa/3.0/igo). Under the terms of this licence, you may copy, redistribute and adapt the work for non-commercial purposes, provided the work is appropriately cited, as indicated below. In any use of this work, there should be no suggestion that WHO endorses any specific organization, products or services. The use of the WHO logo is not permitted. If you adapt the work, then you must license your work under the same or equivalent Creative Commons licence. If you create a translation of this work, you should add the following disclaimer along with the suggested citation: “This translation was not created by the World Health Organization (WHO). WHO is not responsible for the content or accuracy of this translation. The original English edition shall be the binding and authentic edition”. Any mediation relating to disputes arising under the licence shall be conducted in accordance with the mediation rules of the World Intellectual Property Organization. (http://www.wipo.int/amc/en/mediation/rules/) Suggested citation. Haigh K, Twabi H, Ryan H, Lutje V, Nevitt S, Davies G. Web Annex 2. Optimization of dosage of the first-line medicines rifampicin, isoniazid, ethambutol and pyrazinamide in treatment of drug-susceptible tuberculosis: summary of evidence from four systematic reviews. In: WHO operational handbook on tuberculosis. Module 4: treatment - drug-susceptible tuberculosis treatment. Geneva: World Health Organization; 2022. Licence: CC BY-NC-SA 3.0 IGO. Cataloguing-in-Publication (CIP) data. CIP data are available at http://apps.who.int/iris. Sales, rights and licensing. To purchase WHO publications, see http://apps.who.int/bookorders. To submit requests for commercial use and queries on rights and licensing, see https://www.who.int/copyright. Third-party materials. If you wish to reuse material from this work that is attributed to a third party, such as tables, figures or images, it is your responsibility to determine whether permission is needed for that reuse and to obtain permission from the copyright holder. The risk of claims resulting from infringement of any third-party-owned component in the work rests solely with the user. General disclaimers. The designations employed and the presentation of the material in this publication do not imply the expression of any opinion whatsoever on the part of WHO concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. Dotted and dashed lines on maps represent approximate border lines for which there may not yet be full agreement. The mention of specific companies or of certain manufacturers’ products does not imply that they are endorsed or recommended by WHO in preference to others of a similar nature that are not mentioned. Errors and omissions excepted, the names of proprietary products are distinguished by initial capital letters. All reasonable precautions have been taken by WHO to verify the information contained in this publication. However, the published material is being distributed without warranty of any kind, either expressed or implied. The responsibility for the interpretation and use of the material lies with the reader. In no event shall WHO be liable for damages arising from its use. The named authors alone are responsible for the views expressed in this publication. This publication forms part of the WHO document entitled WHO operational handbook on tuberculosis. Module 4: treatment - drug-susceptible tuberculosis treatment. It is being made publicly available for transparency purposes and information. Contents Acknowledgements ............................................................................................... 3 Background ............................................................................................................ 4 Objectives ..............................................................................................................4 Methods .................................................................................................................4 Key findings ............................................................................................................5 Conclusions ........................................................................................................... 7 Annex 1 Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review 1. Background ..................................................................................................... 12 2. Results ............................................................................................................ 20 3. Discussion ....................................................................................................... 32 Appendix 1. Search strategy ............................................................................... 39 Appendix 2. Characteristics of the studies included .......................................... 41 Appendix 3. Studies that were excluded ............................................................ 51 Appendix 4. Data and analyses ........................................................................... 52 Annex 2 Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review 1. Background ...................................................................................................... 64 2. Results ............................................................................................................. 72 3. Discussion ........................................................................................................ 82 Appendix 1. Search strategy ............................................................................... 89 Appendix 2. Characteristics of the studies included .......................................... 91 Appendix 3. Studies that were excluded from the review ................................ 119 Appendix 4. Funnel and Galbraith plots ............................................................ 121 Appendix 5. Data and analyses ......................................................................... 123 Appendix 6. Summary of findings ..................................................................... 129 Optimization of dosage of the first-line medicines rifampicin, isoniazid, ethambutol and pyrazinamide in treatment of drug-susceptible tuberculosis: summary of evidence from four systematic reviews Summary report2 Annex 3 Optimization of the dose of ethambutol in adults with presumed drug-susceptible tuberculosis: systematic review 1. Background .................................................................................................... 137 2. Results ........................................................................................................... 145 3. Conclusions ................................................................................................... 147 4. Authors’ conclusions ..................................................................................... 149 5. References ..................................................................................................... 150 Appendix 1. Search strategy ............................................................................. 152 Appendix 2. Excluded Studies .......................................................................... 153 Annex 4 Optimization of the dose of pyrazinamide in adults with presumed drug-susceptible tuberculosis: systematic review 1. Background .................................................................................................... 158 2. Results ........................................................................................................... 166 3. Discussion ...................................................................................................... 172 Appendix 1. Search strategy ............................................................................. 177 Appendix 2. Studies that were excluded .......................................................... 179 Appendix 3. Summary of findings ..................................................................... 180 Optimization of dosage of the first-line medicines rifampicin, isoniazid, ethambutol and pyrazinamide in treatment of drug-susceptible tuberculosis: summary of evidence from four systematic reviews Summary report 3 Acknowledgements The production and writing of this document – Optimization of dosage of the first-line medicines rifampicin, isoniazid, ethambutol and pyrazinamide in treatment of drug-susceptible tuberculosis: summary of evidence from four systematic reviews – was conducted by the University of Liverpool and the groups of experts: Kathryn Haigh, Hussein Twabi, Hannah Ryan, Vittoria Lutje, Sarah Nevitt with the overall guidance of Dr. Gerry Davies. The work was coordinated by the WHO Global TB Programme. Optimization of dosage of the first-line medicines rifampicin, isoniazid, ethambutol and pyrazinamide in treatment of drug-susceptible tuberculosis: summary of evidence from four systematic reviews Summary report4 Background All four drugs comprising the current first-line anti-tuberculosis (TB) regimen (rifampicin (RIF), isoniazid (INH), pyrazinamide (PZA), ethambutol (EMB)) were developed before 1970, when drug development programmes were very different from those of today. In particular, justification and exploration of dosing strategies was often informal and incomplete, relying on subsequent post-marketing studies to determine the risk–benefit ratio of the dose levels used in practice. Recently, however, the currently accepted doses of the four drugs have been re-evaluated according to modern preclinical and clinical pharmacokinetics and pharmacodynamics and, in some cases, new early-phase clinical trials to establish the proof-of-concept that higher doses could safely improve treatment outcomes. It has not, however, been established whether the possible benefits will be realized in terms of long-term outcomes and without important additional toxicity. Maximization of the efficacy of the first-line regimen could be important in improving long-term outcomes of TB treatment for all patients and making it more robust to variations in adherence, pharmacokinetics and pharmacogenetics and to the emergence of resistance. Furthermore, intensification of treatment with higher doses could be important for people with severe or disseminated disease and in populations who are particularly vulnerable to increased variation in pharmacokinetics or drug–drug interactions. Objectives To assess the efficacy and safety of doses of first-line TB drugs (RIF, INH, PZA ,EMB) higher than those currently recommended by the World Health Organization (WHO) when used as part of a combination regimen for treating or re-treating people with presumed drug-susceptible TB. Methods We conducted four systematic reviews, one for each of the first-line drugs, according to a single master protocol. We included randomized controlled trials of adults being treated for the first time for pulmonary or extrapulmonary TB or being re-treated after a previous episode. We included trials that included direct comparisons of currently recommended doses with higher doses of any of the first-line drugs in otherwise identical background regimens (in terms of companion drugs and duration of treatment). Doses were expressed as the actual weight-adjusted dose, and regimens with less-than-daily scheduled dosing were excluded. We extracted outcomes that conformed as closely as possible to programmatic definitions, including treatment success, treatment failure, recurrence, death, total adverse events, serious adverse events, treatment discontinuation and drug-specific adverse events of interest. We used a highly sensitive strategy to identify trials, and the review process conformed to the PRISMA principles and PRISMA-P checklist. Abstracts and full texts were screened and data extracted by two authors onto a pilot- tested form. We used the Cochrane Risk of Bias tool to assess study quality. We expressed effects as risk ratios and risk differences calculated from intention-to-treat, per-protocol and appropriate sensitivity analyses. We performed meta-analyses with a random effects model and expressed heterogeneity with the I2 statistic. We planned but could not perform meta-regression Optimization of dosage of the first-line medicines rifampicin, isoniazid, ethambutol and pyrazinamide in treatment of drug-susceptible tuberculosis: summary of evidence from four systematic reviews Summary report 5 or network meta-analyses because of limitations of the available data. We assessed the certainty of the evidence with the GRADE approach and calculations of the optimal information size. Key findings Rifampicin (RIF) Of the 13 trials (4411 participants) included, 8 were in patients with pulmonary TB, 4 in patients with TB meningitis and one on all forms of TB. The doses of RIF ranged from 10 mg/kg to 35 mg/kg, but doses up to 20 mg/kg were used in the majority of the trials. The studies included were mostly early-phase trials with small samples, well below the optimal information size for determining efficacy outcomes of interest. Five were double-blinded, while in two others the assessors were blinded for efficacy but not for safety outcomes. There was no evidence of higher rates of treatment success at RIF doses of 15 mg/kg (RR 1.00, [0.97–1.03], RD –1.0% [–5.5%– 3.5%], 4 trials, 916 participants), 20 mg/kg (RR 1.01, [0.97–1.06], RD 2.9% [–1.1%–7.0%], 6 trials, 1302 participants) or 35 mg/kg (RR 0.98, [0.91–1.06], 1 trial, 186 participants). The evidence suggested that RIF doses of 15 mg/kg may reduce treatment failure (RR 0.71 [0.27– 1.88], RD–1.0% [–3.6–1.6%], 2 trials, 616 participants), while a dose of 20 mg/kg may result in no difference (RR 1.01 [0.29–3.61], RD 0.5% [–1.9–2.8%], 2 trials, 821 participants). RIF doses of 15 mg/kg (RR 0.79 [0.25–2.45], RD –1.6% [–4.1–0.7%], 2 trials, 617 participants) or 20 mg/ kg (RR 0.93 [0.07–12.72], RD 0.0% [–1.2–1.2%], 2 trials, 821 participants) may slightly reduce the risk of relapse. The evidence suggested that RIF doses of 15 mg/kg (RR 0.80 [0.47–1.37], RD 1.1% [–4.0–1.8%], 4 trials, 916 participants) or 20 mg/kg (RR 0.93 [0.47–1.81], RD 0.5% [–2.2–1.3%] 7 trials, 1341 participants) probably slightly reduce all-cause mortality. For RIF at doses of 20 mg/kg (RR 0.96 [0.53–1.76], RD 0.6% [–3.0–1.7%], 6 trials, 1 310 participants), the evidence suggested little or no difference in the number of serious adverse events, while for RIF doses of 15 mg/kg (RR 1.83 [0.5–6.69], RD 1.4% [–1.4%–4.2%]. 3 trials, 417 participants), it was suggested that higher doses may increase the number of serious adverse events, but the evidence is very uncertain. There was little or no difference in risk of drug-induced liver injury at an RIF dose of 15 mg/kg (RR 0.99 [0.54–1.83], RD 1% [–2.8–5.0%], 3 trials, 420 participants), but the evidence was very uncertain, while for RIF at 20 mg/kg (RR 0.99 [0.63–1.56], RD 0.0% [–2.3–2.3%], 6 trials, 1310 participants) the evidence suggested little or no difference. For doses of 30–35 mg/kg, the evidence for serious adverse events and drug-induced liver injury was very uncertain. In patients with TB meningitis, two studies evaluated neurological recovery. The larger trial concluded that there was probably little or no difference in this outcome with a dose of RIF of 15 mg/kg. Isoniazid (INH) Four trials (901 participants) were included, all in patients with pulmonary TB. The doses of INH ranged from 3 to 12 mg/kg, but a different direct dose comparison was used in each trial, and neither conventional nor network meta-analysis could be performed. All the trials included were judged to be at high risk of bias and were well below the optimal information size for the majority of outcomes. Furthermore, all were published before 1969. Although treatment success probably improved when the dose of INH was increased from 4 to 6 mg/kg, the rate was numerically or significantly lower in all the trials that evaluated higher doses, in the range 9–12 mg/kg. Similarly, Optimization of dosage of the first-line medicines rifampicin, isoniazid, ethambutol and pyrazinamide in treatment of drug-susceptible tuberculosis: summary of evidence from four systematic reviews Summary report6 with the exception of one trial, the occurrence of treatment failure was probably similar at higher doses of INH, and none of the trials reported on the relapse outcome. We found some evidence of increasing numbers of adverse events with higher doses of INH. There were more adverse events and/or serious adverse events in all the studies included, particularly in those in which the highest doses were used. Pyrazinamide (PZA) One trial (628 participants) in patients with pulmonary TB was included. The trial was published in 1959. Doses of 25 and 40 mg/kg were compared with two background regimens of INH and para-aminosalicylic acid (PAS) and reported only safety outcomes. The methods used to evaluate drug-induced liver injury were unclear. The trial was considered to be at high risk of bias and the evidence very uncertain. Higher doses of PZA might have been associated with higher risks of discontinuation (RR 2.39 [0.79–7.28], RD 4% [1–8%], 1 trial, 628 participants), drug-induced liver injury (RR 1.65 [0.49–5.62], RD 2% [–3–7%], 1 trial, 628 participants) and death (RR 1.17 [0.55–2.49], RD 1% [–3–5%], 1 trial, 628 participants), but the associations varied according to the background regimen. Ethambutol (EMB) No trials met the inclusion criteria for the review. Optimization of dosage of the first-line medicines rifampicin, isoniazid, ethambutol and pyrazinamide in treatment of drug-susceptible tuberculosis: summary of evidence from four systematic reviews Summary report 7 Conclusions Our review addressed programmatically important long-term efficacy and safety outcomes; however, the available evidence was derived from early-phase trials in which these outcomes were not the primary ones and from trials conducted before current standards were set in which the companion drugs were different from those in the current first-line regimen. There was concern about high risks of bias in some of the older trials and about lack of blinding with respect to safety outcomes in more recent trials. We were unable to analyse indirect comparisons or formal dose–response relations because of limitations in the data. There was no evidence that RIF at doses up to 20 mg/kg increased the rate of treatment success, although it might have slightly reduced the rate of relapse. We found no convincing evidence of increased risks for adverse events or drug-induced liver injury in this dose range. The evidence for doses of 30–35 mg/kg was very uncertain for both efficacy and safety outcomes. Similarly, we found no consistent evidence of dose–response relations in efficacy outcomes for INH at doses up to 12 mg/kg and some evidence that the risk of adverse events was increased. The evidence for higher doses of PZA, up to 40 mg/kg, was very uncertain, but these doses may have been associated with higher rates of drug-induced liver injury and discontinuation. We found no trials of direct comparisons of doses of EMB. Our review therefore does not provide evidence for adjustment of the dose of any of the drugs used in first-line regimens, and we were unable to identify important subgroups who might benefit from increased doses. For detailed systematic reviews and meta-analyses for RIF, INH, EMB and PZA, see annexes 1, 2, 3 and 4, respectively.

Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review ANNEX 1 Rebecca Crook1, Francesca Liuzzi1, Hannah Ryan2, Sarah Nevitt2, Geraint Davies2 1 Royal Liverpool University Hospital, United Kingdom 2 University of Liverpool, United Kingdom

Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review Annex 1 11 Contents 1. Background 12 1.1 Description of the condition 12 1.2 Description of the intervention 12 1.3 How the intervention might work 13 1.4 Why this review is important 13 1.5 Methods 14 2. Results 20 2.1 Description of studies 20 2.2 Effects of interventions 25 3. Discussion 32 3.1 Summary of main results 32 3.2 Overall completeness and applicability of evidence 33 3.3 Certainty of the evidence 34 3.4 Potential biases in the review process 34 3.5 Agreements and disagreements with other studies and reviews 35 3.6 Authors’ conclusions 35 3.7 Acknowledgements 36 3.8 References 36 Appendix 1. Search strategy 39 Appendix 2. Characteristics of the studies included 41 Appendix 3. Studies that were excluded 51 Appendix 4. Data and analyses 52 Analysis 1. Treatment success, intention-to-treat analysis of relative risk of low dose vs high dose INH 52 Analysis 2. Treatment success, per protocol analysis of relative risk of low dose vs high dose INH 53 Analysis 3. Treatment failure, intention-to-treat analysis of relative risk of low dose vs high dose INH 53 Analysis 4. Treatment failure, per-protocol analysis of relative risk of low dose vs high dose INH 54 Analysis 5. All-cause deaths, intention-to-treat analysis of relative risk of low dose vs high dose INH 54 Analysis 6. All causes of death, per-protocol analysis of relative risk of low dose vs high dose INH 55 Analysis 7. TB-related death, intention-to-treat analysis of relative risk of low dose vs high dose INH 55 Analysis 8. TB-related deaths, per-protocol analysis of relative risk of low dose vs high dose INH 56 Analysis 9. Total adverse events, intention-to-treat analysis of relative risk of low dose vs high dose INH 56 Analysis 10. Serious adverse events, intention-to-treat analysis of relative risk of low dose vs high dose INH 57 Analysis 11. Adverse events requiring drug discontinuation, intention-to-treat analysis of relative risk of low dose vs high dose INH 57 Analysis 12. Drug-induced liver injury, intention-to-treat analysis of relative risk of low dose vs high dose INH 58 Analysis 13. Drug-induced peripheral neuropathy, intention-to-treat analysis of relative risk of low dose vs high dose INH 58 Analysis 14. Other adverse events, intention-to-treat analysis of relative risk of low dose vs high dose INH 59 Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review Annex 112 1. Background 1.1 Description of the condition Tuberculosis (TB) remains the single biggest killer of all infectious diseases, with an estimated 10 million cases and 1.2 million deaths worldwide in 2018 (1). While 30 low- and middle-income countries account for almost 90% of this burden, a significant fraction can be attributed to risk factors such as HIV co-infection (9%) and diabetes (15%). While progress has recently been made in the control of TB, the annual decrease in incidence does not currently exceed 2%. Seven million people were treated with first-line TB regimens globally in 2018, with a treatment success rate of 85%, but estimates suggest that the case fatality rate among all incident cases was as high as 15%, with most of the deaths among people in the most economically active period of their lives (1). Although pulmonary disease is the commonest presentation of TB, extrapulmonary and disseminated disease, particularly with neurological involvement, may be associated with worse outcomes. In addition, some people with TB harbour resistant strains of Mycobacterium tuberculosis for which first-line therapy is not effective (2). Of those presenting with TB for the first time, 13.1% are resistant to isoniazid (INH) and 3.3% are resistant to rifampicin (RIF) and require modified, longer, more expensive treatment regimens (1). 1.2 Description of the intervention First-line short-course regimens for TB have changed little since their introduction more than 40 years ago. The finding that when RIF and pyrazinamide (PZA) were added to INH the duration of therapy could be reduced to as little as 6 months established this trio of drugs as the backbone of effective therapy (3). A fourth drug was often used to prevent the emergence of resistance, particularly in people who already harboured INH-resistant strains. While streptomycin was initially widely used for this purpose, widespread resistance and need for parenteral administration led to its replacement by the oral agent ethambutol (ETH), resulting in the current standard first- line regimen (4). INH is a synthetic analogue of nicotinamide, first introduced for the treatment of TB in 1952. It inhibits mycolic acid synthesis and a pro-drug activated by the mycobacterial catalase- peroxidase katG (5). INH greatly improved the results of early combination TB regimens containing streptomycin and para-aminosalicylic acid (PAS) (6) and quickly became established as an indispensable element of all combination regimens. Because of its widespread early use as monotherapy or as part of relatively ineffective regimens, however, resistance to INH subsequently spread rapidly worldwide. INH monoresistance is associated with a three times higher risk of failure of the current first-line regimen (2). The currently recommended daily dose of INH for adults is 5 (4–6) mg/kg, adjusted for weight according to different banding schedules that depend on formulation as a single drug or as part of a fixed-dose combination (7). For intermittent administration, doses as high as 15 mg/kg have been recommended, while 20 mg/kg has been used in neurological and drug-resistant disease (8). INH is metabolized primarily through N-acetylation, a highly polymorphic metabolic pathway, controlled by the NAT2 locus. This results in distinct fast, intermediate and slow metabolizer phenotypes, which vary geographically, and differences in plasma area-under-the-curve (AUC) of up to five times between fast and slow acetylators (9). Current dosing strategies do not take Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review Annex 1 13 this variability into account. The two most important toxic effects of INH, drug-induced liver injury and peripheral neuropathy, have been linked to acetylator phenotype and plasma exposure (10). The latter is due to inhibition of pyridoxal phosphate-dependent enzyme systems and can be averted by supplementation with pyridoxine (11). The dose and concentration relations of INH have been extensively evaluated in early bactericidal activity (EBA) studies. EBA increases monotonically over a range of daily doses from 10 to 300 mg, although it appears to plateau at 600 mg (12, 13). This relation also appears to hold in TB patients harbouring organisms with low-level resistance mediated by inhA mutations at doses up to 15 mg/kg (14). For typical patients with fully susceptible organisms, however, considerations of risk and benefit based on available dose–response and dose–toxicity data have thus favoured current dosing recommendations. Whether certain subgroups may benefit from higher doses has been incompletely explored. Modern first-line regimens rely on weight-based dosing strategies to control variations in pharmacokinetics and, after prolonged pharmaceutical development, are co-formulated as fixed-dose combination tablets by a number of manufacturers (15). While truly individualized dosing approaches have been advocated, with therapeutic drug monitoring to identify individuals at the lower end of the distribution of pharmacokinetics, such strategies have not been widely implemented, even in settings with the necessary technical resources. For national TB programmes globally, altered dosing recommendations for all TB patients or for selected subgroups are the only interventions that could be immediately implemented worldwide. For that reason, this review focuses on evidence suitable for universal or stratified rather than individualized dosing recommendations. 1.3 How the intervention might work INH is among the most active bactericidal drugs against replicating M. tuberculosis in many preclinical systems and has the highest EBA in early-phase clinical trials during the first few days of treatment. It therefore has a key early role in rapidly reducing the burden of organisms and preventing the acquisition of resistance (16). Although doubts are frequently expressed about its activity against “persister” organisms as treatment progresses, INH is effective in eliminating latent TB infection (17) and greatly improves long-term outcomes in early combination regimens for pulmonary TB (18). While the EBA of INH appears to be maximized at the current dose of 300 mg (19), other studies have emphasized the important of pharmacogenetic variability in exposure to the drug due to control of its polymorphic metabolism by the NAT2 locus (20). For this reason, dose optimization has been suggested to mitigate against sub-optimal responses in patients with the fast acetylator phenotype (21). In addition, higher doses of INH could improve treatment outcomes in patients with organisms that harbour inhA mutations by optimizing AUC/MIC targets, addressing one of the commonest forms of resistance and causes of treatment failure on first-line regimens worldwide. Whether these gains in efficacy can be achieved without clinically significant increases in important toxic effects such as drug-induced liver injury is unknown. 1.4 Why this review is important Several lines of evidence suggest that strategies for dosing the individual drugs that comprise the current first-line regimen for TB may not be optimal. Maximizing the efficacy of the regimen Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review Annex 114 could be important in improving the long-term outcomes of TB treatment for all patients and making it more robust to variations in adherence, pharmacokinetics and pharmacogenetics and to the emergence of resistance. Furthermore, intensification of treatment with higher doses could be important for people with severe or disseminated disease and in populations particularly vulnerable to increased variation in pharmacokinetics or drug–drug interactions. Objective To assess the efficacy and safety of doses of INH higher than those currently recommended by WHO when used as part of a combination regimen for treating people with presumed drug- susceptible TB. 1.5 Methods The review was conducted according to the principles outlined in the PRISMA statement and PRISMA-P checklist (22, 23). Criteria for considering studies for this review Types of studies Randomized controlled trials Types of participants Inclusion criteria • Adults being treated for TB for the first time or being re-treated after a previous episode of TB Exclusion criteria • People being treated for TB with confirmed resistance to RIF, INH, PZA or ETH All forms of TB (pulmonary, extrapulmonary and disseminated) were considered for the purposes of the review, whether the diagnosis was based on bacteriological confirmation, imaging, biopsy or a clinical decision to treat. Types of interventions Intervention • Anti-TB treatment regimens containing INH at doses higher than those recommended in current WHO guidelines Comparator • Anti-TB treatment regimens containing INH at doses recommended in current WHO guidelines Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review Annex 1 15 The intervention was defined on the basis of comparisons based on dose, not on target concentrations in plasma, and randomized concentration-controlled trials were not included in the review. Although not recommended by WHO, administration of INH by the intravenous route as an alternative to oral was permissible, provided that the target dose was different from those currently recommended. The dose metric used was actual weight-adjusted doses expressed in mg/kg. It was expected that many studies would not specify the target weight-adjusted dose in the report. When absolute doses were specified, the weight-adjusted dose for each trial was determined when possible from the description of weight data provided in the study report or, if these data were not available, the average weight in all the trials included in the analysis. When weight-banded dosing regimens were used, the average target weight-adjusted dose stated in the study report or the manufacturer’s summary of product characteristics was accepted, or, if not specified, the average of weight-adjusted doses computed from the mid-point of each band. Comparisons according to dose for each drug were conducted between regimens of similar or different backbone chemotherapy compositions and duration. Direct comparisons between regimens of different doses and durations were not considered in the primary analysis. Types of outcome measures Primary outcomes We used definitions consistent with WHO programmatic outcomes (24), adapted to permit evaluation of regimens longer or shorter than 6 months, of follow-up data in clinical trials and of trials of extrapulmonary TB. Treatment success In adults with pulmonary TB (bacteriologically confirmed, presumed drug-sensitive): smear- or culture-negative in the last month of treatment and on at least one previous occasion OR completed treatment without evidence of failure BUT with no record to show that sputum smear or culture results in the last month of treatment and on at least one previous occasion were negative, either because tests were not done or because results are unavailable. In adults with extrapulmonary TB: resolution of clinical signs and symptoms of TB at the end of treatment as judged by the investigators Treatment failure In adults with pulmonary TB (bacteriologically confirmed, presumed drug-sensitive): sputum smear or culture positive within the last month of treatment. In adults with extrapulmonary TB: failure to resolve or return of clinical signs and symptoms of TB by the end of treatment as judged by the investigators Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review Annex 116 Relapse In adults with pulmonary TB (bacteriologically confirmed, presumed drug-sensitive): smear or culture positive on one or more occasions during a defined period of follow-up after having previously met the criteria for treatment success. In adults with extrapulmonary TB: return of clinical signs and symptoms of TB as judged by the investigators after having previously met the criteria for treatment success Death Death from any cause before starting, during the course of treatment or during follow-up. Adverse events All reported adverse events expressed as: total number of events and/or number of participants experiencing an event Secondary outcomes Serious adverse events All adverse events detailed in the study report that resulted in death, were life-threatening, requires hospitalization or prolongation of hospitalization, resulted in persistent or significant disability or incapacity, consisted of a congenital anomaly or birth defect or were another important (protocol- defined) medical condition. Discontinuation Adverse events resulting in discontinuation of study medication Drug-specific adverse events of interest Drug-induced liver injury Peripheral neuropathy Disease-specific efficacy outcomes of interest Outcomes specific to trials in which people with extrapulmonary forms of TB were recruited, such as neurological disability in meningeal TB and pericardial constriction in pericardial TB. Search methods for identification of studies We attempted to identify all relevant studies, regardless of language, date of publication or publication status (published, unpublished, in press, in progress). Electronic searches We searched using the following electronic resources: MEDLINE, EMBASE, CENTRAL (Cochrane central register of controlled trials), Cochrane Infectious Diseases Group clinical trials register, WHO International Clinical Trials Registry and Clinicaltrials.gov. We used the search strategy shown in Annex 1 (for CENTRAL, MEDLINE and EMBASE), which consists of the Cochrane Highly Sensitive Search Strategies for identification of clinical trials appropriate to those resources (25). Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review Annex 1 17 Search for other resources We searched the reference lists of the retrieved study reports for any other studies. We also cross-referenced the results of the searches to the database of clinical trials of first-line TB treatment generated by our group for the systematic reviews conducted by the PreDiCT-TB consortium (6, 18, 26). Data collection and analysis Selection of studies After de-duplication, the titles and abstracts of unique studies identified in the search were screened independently by two authors against the inclusion and exclusion criteria, and the full text was retrieved for all citations considered “eligible” or “unclear” but not for those considered “ineligible”. Two authors independently screened the retrieved full-text study reports, marked them as “eligible” or “ineligible” and recorded the reasons for exclusion in each case. Any disagreements were resolved by discussion with a third author. We used the Covidence web- based interface (27) to manage these processes and summarized the selection process in a PRISMA flow diagram. Data extraction and management We designed and pilot-tested a data extraction form on a small subset of studies and modified it according to the results. After the form format was finalized and programmed in Covidence, two authors independently extracted data and compared the results. Any errors or discrepancies were resolved by discussion and comparison with the study reports. We extracted the following data from each included study, when available: Source: Lead author, year of publication, journal, PubMed ID, sponsor, trial registration number, corresponding author Methods: Study design, study dates and duration, sequence generation, allocation concealment, blinding, completeness of outcome data, selective reporting, other concerns about bias Participants: Number recruited, country, patient type (new/retreatment), setting, inclusion/ exclusion criteria, age, sex, HIV status, diabetes status Microbiological methods: Details of diagnostic and susceptibility testing methods Pharmacology: Availability of data on pharmacokinetics and/or pharmacogenetics Interventions: Number of arms, names of drugs, doses of drugs, frequency of dosing, use of fixed-dose combinations, duration and structure of regimen, use of direct observation Outcomes: For each review outcome, we extracted the number of patients randomized to each arm, the number included in the analysis and the number that experienced the event of interest to enable intention-to-treat (ITT), per-protocol (PP) and best- and worst-case analyses. We recorded the duration of post-treatment follow-up in each case. Assessment of risk of bias Two review authors independently assessed the methodological quality of each included study using the Cochrane “Risk of bias” tool and reported the results in a “Risk of bias” table (28). We Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review Annex 118 resolved any disagreements through discussion. To generate allocation sequence and allocation concealment, we classified each as either adequate, inadequate or unclear for each study according to Jüni et al. (29). We reported who was blinded in each included study and assessed the risk of bias associated with blinding separately for each primary outcome. If at least 90% of participants were followed up to study completion, we classified inclusion of all randomized participants as adequate; otherwise, we classified inclusion as inadequate. We attempted to contact the study authors for information that was unspecified or unclear. Measures of treatment effect We used the risk ratio (RR) as the measure of treatment effect for analyses of both efficacy and safety and also expressed the results as the risk difference (RD). Missing data The primary outcomes were analysed on both an ITT and a PP basis to account for the effect of missing data. Assessment of heterogeneity We assessed heterogeneity by visual inspection of forest plots to determine the closeness of point estimates to each other and overlap of confidence intervals (CIs). For direct comparisons and meta-regression analyses, we used the chi² test with a significance level of 0.10 (30) and the I² statistic (31) to assess heterogeneity, with a value of 50% taken to indicate significant statistical heterogeneity. For indirect comparisons, the τ2 statistic was also used to assess heterogeneity. Assessment of reporting biases We planned to visually inspect the funnel and Galbraith plots for the studies for any evidence of publication bias if at least 10 studies were included in the analysis. As we included only four studies in our review, we were unable to do so. Data synthesis We prioritized direct within-trial comparisons between doses of the individual drugs that were part of otherwise similar regimens at the same total duration of treatment, to the extent permitted by the data. We were unable to perform meta-regression or form a feasible network of comparisons because of the small number of trials and diverse background regimens. Meta-analysis was performed with random effects models with the DerSimonian–Laird method (32). Stratified and unstratified forest plots were used to present the data graphically. Corresponding 95% CIs and P values will be computed, with a significance level of 0.05. We quantified heterogeneity with the I2 statistic. Statistical analyses were conducted in R 3.4 (33) with the package metafor. Subgroup analysis and investigation of heterogeneity For each of the drugs, we planned to stratify the primary and secondary analyses specified above by the subgroups listed below when the data permitted meaningful characterization and analysis. The four studies included did not, however, allow such stratification. • People receiving treatment for the first time versus those previously treated • Pulmonary versus specific forms of extrapulmonary TB (for example meningeal TB) • Severity of disease (as measured by smear grade, radiological extent of disease, presence of cavitation, disseminated disease and/or mycobacteraemia) • HIV co-infection Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review Annex 1 19 • Diabetes • Daily versus less-than-daily dosing schedule of the regimen (two or three times per week) • Weight-banded versus non-weight banded dosing strategies • Companion drugs in the regimen • Pharmacogenetic polymorphisms (phenotypic or genotypic acetylator status) We were also unable to conduct subgroup analyses because of small numbers for each of the following outcomes in each study: total deaths, serious adverse events, drug-induced liver injury, peripheral neuropathy and other adverse events. Certainty of the evidence We assessed the certainty of the evidence with the GRADE approach (34). We had planned to use GRADEpro GDT (35) to construct a “summary of findings” table presenting ratings of the certainty of evidence for effect estimates for each outcome with relative and absolute measures of effect. Because we could not combine data at common dose levels for any outcome and because the certainty of evidence from the trials was rated overall as very low, we did not do this. Sensitivity analysis We performed an ITT analysis, which included all participants who were randomized, and a PP analysis, which included only participants who were randomized and completed the treatment course as planned. Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review Annex 120 2. Results 2.1 Description of studies Results of the search The search identified 4245 records, of which 1788 were identified as duplicates. Of the remaining 2457, we excluded 2350 after assessing the titles and abstracts, leaving 107 full-text publications for assessment. Fig. 1 shows the screening process. Fig. 1. PRISMA diagram Search N=4245 Title/Abstract Screening N=2457 Included in Qualitative Synthesis N=4 Included in Quantitative Synthesis N=1 Full-text Retrieval N=107 Eligible INH N=13 Eligible RIF, INH, PZA or ETH N=38 Duplicates N=1788 Excluded N=2350 Excluded N=69 33 wrong intervention 30 wrong comparator 3 wrong study design 2 wrong population 1 duplicate Excluded N=9 3 wrong intervention 2 wrong population 2 not retrievable 1 no extractable data 1 duplicate Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review Annex 1 21 Included studies Four trials were included with a total of 901 participants. Annex 2 provides full details of each trial. Geographical location and period Two studies (36, 37) recruited patients in Kenya, Uganda and the United Republic of Tanzania; a third was conducted in Switzerland (38) and the fourth in the USA (39). At the time of writing, Kenya and Uganda were considered to be high-burden TB countries by WHO (40) and low-income countries by the World Bank (41). The United Republic of Tanzania, a low- to middle-income country, Switzerland and the USA (both high-income countries) are not on the WHO high-burden list (40, 41). None of the countries is considered to have a high burden of multi-drug-resistant or RIF-resistant TB according to WHO definitions (42). Patients were recruited into the studies between 1953 and 1968. Participants All four studies included participants with pulmonary TB who had not received significant previous treatment and who can therefore be considered new cases. The study conducted by the East African and British Medical Research Councils in 1963 (EA/BMRC 1963) fully described baseline susceptibility to the trial medication and included 5 participants resistant to PAS and 10 resistant to thiacetazone out of a total of 278 tested participants; 40 patients with INH resistance were included, but the results were presented in a separate publication. The study conducted by EA/BMRC in 1966 (EA/BMRC 1966) included 20 patients with INH resistance, also described in a separate publication; the authors did not fully report rates of resistance to the background drugs in the regimen (thiacetazone and PAS). Favez et al. (38) excluded all 10 participants with baseline resistance to its trial medication, while the US Public Health Service Cooperative (USPHS) did not report resistance testing, and the participants could be categorized only as presumed drug sensitive. EA/BMRC 1963 and 1966 both recruited patients aged ≥ 15 years, while Favez et al. and the USPHS 1954 did not report age criteria. All age data were collected categorically, with median age categories reported as 24–34 years, < 35 years, 31–40 years and 33–54 years in the four trials, respectively. HIV testing and use of antiretroviral therapy were not reported in any of the studies. Diabetes screening was reported only in EA/BMRC 1963, in which one patient with diabetes was excluded from the study. Interventions The doses of INH given to the comparator groups in each trial ranged from 3 to 5 mg/kg, and the intervention doses ranged from 6 to 12 mg/kg. While EA/BMRC 1963 and USPHS 1954 each had four arms, these contributed only two direct comparisons because of variation in the backbone regimen between each of the arms. All the comparisons in this review are based on daily dosing of INH. Half of the studies – EA/BMRC 1963 and 1966 – did not specify the target weight-adjusted dose. This was calculated for both studies from the mean weight reported in EA/BMRC 1966, as EA/BMRC presented weight as a categorical variable with incompletely described limits on Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review Annex 122 the banding, thereby making a mean calculation invalid. No weight-banded dosing regimens were used in any the studies. The duration of treatment in the trials ranged from 10 (USPHS 1954) to 18 months (EABMRC 1966). EA/BMRC 1963 reported that patients were eligible for a change in therapy change at 6 months if they remained sputum smear-positive at that tune. Favez et al. did not specify the intended duration of treatment. All the trials were published well before introduction of the modern first-line regimen containing RIF and PZA (43). The backbone regimens included oral thiacetazone, PAS, thiocarlide and streptomycin. Other interventions described include routine administration of pyridoxine to all patients by Favez et al. In addition, EA/BMRC 1966 included secondary randomization of all participants to two further interventions: (i) complete outpatient care or a 2-month initial period of inpatient care and (ii) addition of two surprise home visits or no such visits. Participants were hospitalized for the duration of treatment in the studies of Favez et al. and USPHS, while those in EA/BMRC 1963 were treated as inpatients for the first 6 months and as outpatients for the second 6 months. EA/BMRC 1966 divided participants equally into two groups, one receiving outpatient care throughout and the other receiving an initial 2 months of inpatient care before continuing with ambulatory care. Follow-up All the trials except Favez et al. followed up participants from treatment onset to completion. Favez et al. reported bacteriological follow-up to 5 months and radiological follow-up to 7 months, it is unclear whether the latter time represents the end of treatment or simply the end of data collection. Follow-up was monthly in all the trials. Outcome measures All trials reported treatment success and failure. None defined them as per the review protocol, and the authors’ definitions were accepted (see Annex 2). Only EA/BMRC 1963 presented outcomes at the end of treatment, while EA/BMRC 1966 reported at 12 months of an 18-month treatment programme and USPHS 1953 reported at 20 weeks of a 40-week treatment programme. Favez et al. reported all outcomes at 5 months (total treatment duration not documented). None of the trials reported on relapse after treatment cessation. Deaths were fully reported in two studies (EA/BMRC 1963 and 1966) and TB-related deaths alone in USPHS 1954. Favez et al. did not report deaths. EA/BMRC 1963 and Favez et al. reported all adverse events; however, the latter attributed all of them to streptomycin. EA/BMRC 1966 reported the numbers of adverse events and serious adverse events, but it did not clearly indicate in which group participants who experienced drug- induced liver injury were. USPHS 1954 reported only adverse events requiring discontinuation of treatment. Excluded studies Of the 13 abstracts included for full paper review, we were unable to obtain the full paper for two. A further seven were excluded after review, three because they used the wrong intervention, two that included the wrong population, one that was a duplicate report and one that was presented so unclearly that data could not meaningfully be extracted. See Annex 3. Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review Annex 1 23 Risk of bias We assessed the risk of bias in all the trials using the Cochrane “Risk of bias” tool (31). See tables in Annex 2. The results of these assessments are summarized in Figs 2 and 3. Fig. 2. Risks of bias in the studies included Fig. 3. Summary of risks of bias in the studies included Sequence generation and allocation concealment EA/BMRC 1963 and 1966 were judged at low risk of bias for sequence generation and allocation concealment, as both describe an adequate method, with a pre-arranged series of sealed envelopes based on random sampling numbers. The risk of bias in these areas in Favez et al. and USHPAS 1954 was unclear, as neither was adequately described. Blinding All the studies were considered at high risk for both performance and detection bias. While it is likely that laboratory personnel may have been blinded to treatment assignment, the only blinding specifically reported was for X-ray interpretation in EA/BRMC 1963 and 1966. As radiological changes were not among our primary or secondary outcomes, the blinding of these assessors did not impact the risk rating in this category. Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review Annex 124 Incomplete outcome data All the studies were considered at high risk for attrition bias due to the high (< 90%) loss of participants to follow-up without specified or clear reasons. EA/BMRC 1966 in particular described a default rate of 71 participants in the first 6 months from the 326 who started treatment. As the treatment arms to which these participants were assigned were not detailed in the trial report, they could not be included in our analysis. EA/BMRC 1963 also presented outcome data for some participants unclearly, such that a reasonable number could not be assigned to their trial arm and their outcomes could not be extracted for analysis (see Annex 2). Selective reporting A published trial protocol could not be identified for any of the trials; however, given their age this was not unexpected and was not factored into the selective reporting risk assessment. EA/BMRC 1963 was deemed to be at unclear risk in this area, as it did not report on non-TB- related deaths in the second 6 months of the trial. EA/BMRC 1966 reported on all expected outcomes and was categorized as at low risk. Favez et al. was assessed as high risk for selective reporting, as it did not report on deaths and reported incompletely on safety end-points. Similarly, USPHS 1954 was considered to be at high risk, as it reported only deaths related to TB and also incompletely described safety end-points. Precision All the trials were small, and meta-analysis was not deemed appropriate. We calculated the optimal information size for the reported outcomes according to ITT event rates in the control arms for a relative effect size of 20% with conventional parameters of α=0.05 and β=0.20. The optimal information size was 288–610 for treatment success, 1676–3276 for treatment failure and larger for the less frequent outcomes of mortality and adverse events. The only outcome for which the certainty of evidence was not downgraded for imprecision was treatment success in EA/BMRC 1966 and USPHS 1954. Other The design of EA/BMRC 1963 included making patients who were sputum smear positive at 6 months eligible for a change of chemotherapy. Of the 300 randomized participants, 228 remained on their allocated treatment for 12 months and 72 did not. The authors, however, presented results at 12 months for all 300 participants by “attempt[ing] to estimate what the response of the patients would have been at 12 months if they had continue[d] on their prescribed regimen throughout the trial”. The way in which these estimates were made is not reported, which introduces a significant potential source of bias. It is also not clear how many of the 72 patients changed chemotherapy because of smear positivity at 6 months and how many for other reasons. It was not possible to extract data exclusively for those who continued on their assigned chemotherapy in order to assess the data accurately assess. Two trials reported outcomes before completion of the planned chemotherapy; EA/BMRC 1966 reported at 12 months of an 18-month regimen, and USPHS 1954 reported most outcomes at 20 weeks and treatment success and failure at 16 weeks, both in a 40-week regimen. The significance to the outcomes of treatment success, treatment failure and deaths is unclear. It is likely, however, that the months of data not reported may have included more adverse events, and these data is likely to be underrepresented in our analysis. Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review Annex 1 25 2.2 Effects of interventions Because of the diversity of INH doses, the duration of the treatment regimens and the timing of outcome data collection, we were unable to conduct an appropriate meta-analysis for any outcome. The only exception was combination of the data for the two comparisons in USPHS 1954 as, except for the INH dose level, they were identical in all respects. The results are therefore presented by study rather than by outcome. The full results of each study are presented in Annex 2 and summary forest plots of each outcome in the analyses in Annex 4. EA/BMRC 1963 EA/BMRC 1963 (153 participants) provided the results of a comparison of doses of 4 mg/kg (published dose, 200 mg) vs 6 mg/kg (published dose, 300 mg). All the results of the trial for the ITT analysis are presented in this order. Treatment success was probably greater with the high dose (6 mg/kg) (56.25% vs 72.60%, RR 1.29 [1.02–1.64], RD 16.35% [1.42–31.28]). Similarly, treatment failure may have been lower at the high dose (6 mg/kg) (28.75% vs 4.11%, RR 0.14 [0.04–0.46], RD –24.64% [–35.56– –13.73]). All-cause mortality might have been higher at the higher dose (3.75% vs 6.85%, RR 1.83 [0.45– 7.38], RD 3.10% [–4.04–10.23]), and the same was true for TB-related mortality (2.50% vs 5.48%, RR 2.19 [0.41–11.61], RD 2.98% [–3.26–9.22]. In both cases, however, the evidence was very uncertain. Total adverse events (7.50% vs 6.85%; RR 0.91 [0.29–2.87], RD –0.65% [–8.83%–7.53%]) and serious adverse events (0.0% vs 1.37%; RR 3.28 [0.14–79.36], RD 1.37 [–1.30–4.04]) might not have differed between dose groups, and this was also true of the risk of drug-induced liver injury and peripheral neuropathy, with small numbers of reported events in each group. For all safety outcomes, the evidence was very uncertain. The results of PP analyses were consistent with those of the ITT analyses. See Table 1 and Annex 4 for full details. Table 1. Findings from EA/BMRC 1963 Intention-to-treat analysis Per-protocol analysis Factor Low dose, 200 mg (4 mg/ kg) High dose, 300 mg (6 mg/ kg) RR (95% CI) RD% (95% CI) Low dose, 200 mg (4 mg/ kg) High dose, 300 mg (6 mg/ kg) RR (95% CI) RD% (95% CI) Participants with data 80 73 – – Varied by outcome Varied by outcome – – Treatment success 45/80 (56.25%) 53/73 (72.60%) 1.29 (1.02– 1.64) 16.35 (1.42 – 31.28) 45/73 (61.6%) 53/59 (89.8%) 1.46 (1.19– 1.78) 28.19 (14.63– 41.75) Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review Annex 126 Intention-to-treat analysis Per-protocol analysis Treatment failure 23/80 (28.75%) 3/73 (4.11%) 0.14 (0.04– 0.46) –24.64 (–35.56– –13.73) 23/73 (31.31%) 3/59 (5.09%) 0.16 (0.05– 0.51) –26.42 (–14.38– –38.46) Relapse Not reported Not reported – – Not reported Not reported – – Death – total 3/80 (3.75% 5/73 (6.85%) 1.83 (0.45– 7.38) 3.10 (–4.04– 10.23) 3/76 (3.95%) 5/66 (7.58%) 1.92 (0.48– 7.73) 3.63 (–4.11– 11.37) Death – TB related 2/80 (2.50%) 4/73 (5.48%) 2.19 (0.41– 11.61) 2.98 (–3.26– 9.22) 2/75 (2.67%) 4/65 (6.15%) 2.31 (0.44– 12.19) 3.49 (–3.40– 10.37) Adverse events – total 6/80 (7.50%) 5/73 (6.85%) 0.91 (0.29– 2.87) –0.65 (–8.83– 7.53) As per ITT analysis Serious adverse events 0/80 (0%) 1/73 (1.37%) 3.28 (0.14– 79.36) 1.37 (–1.30– 4.04) Adverse events requiring dis- continuation of treatment 0/80 (0%) 1/73 (1.37%) 3.28 (0.14– 79.36) 1.37 (–1.30– 4.04) Drug- induced Liver Injury 1/80 (1.25%) 1/73 (1.37%) 1.10 (0.07– 17.20) 0.12 (–3.49– 3.73) Peripheral neuropathy 1/80 (1.25%) 1/73 (1.37%) 1.10 (0.07– 17.20) 0.12 (–3.49– 3.73) Adverse Events – other 4/80 (5.00%) 3/73 (4.11%) 0.82 (0.19– 3.55) –0.89 (–7.49– 5.71) RR, relative risk; RD%, risk difference (%); 95% CI, 95% confidence interval EA/BMRC 1966 EA/BMRC 1966 (251 participants) provided the results of a comparison of 6 mg/kg (published dose, 300 mg) vs 9 mg/kg (published dose, 450 mg). All the results of the trial for the ITT analysis are presented in this order. Treatment success was probably lower at the higher dose (66.14% vs 55.64%, RR, 0.84 [0.69– 1.03], RD –10.50 [–22.50–1.51]), while treatment failure may not have differed between the groups (13.39% vs 19.35%, RR 1.45 [0.82–2.56], RD 5.97% [–3.16%–15.10%]). Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review Annex 1 27 The total number of deaths may not have differed between the two groups (3.15% vs 8.06%, RR 2.56 (0.82–7.95), RD 4.91% [–0.76%–10.60%]). Similarly, the number of deaths related to TB might not have differed between the groups (3.15% vs 6.45%, RR 2.05 [0.63–6.63], RD 3.30% [–1.98%–8.59%]). The total number of adverse events (4.47% vs 6.63%, RR 1.48 [0.62–3.54], RD 2.16 [–2.56– 6.88]) might have been increased at the higher dose, while the number of serious adverse events (2.79% vs 2.76%, RR 0.99 [0.29–3.36], RD –0.03 [–3.43–3.36]) might not have differed between the dose groups. Discontinuation of treatment might also not have differed between the dose groups (2.23% vs 0%, RR 1.76 [0.31–10.19], RD –2.23 [–4.40– –0.07]). The risks of drug-induced liver injury and peripheral neuropathy also appeared to be similar, with small numbers of reported events in each group, but the evidence for all reported safety outcomes was very uncertain. The results of the PP analyses were consistent with those of the ITT analyses. See Table 2 and Annex 4 for full details. Table 2. Findings from EA/BMRC 1966 Intention-to-treat analysis Per-protocol analysis Factor Low dose 300 mg (6 mg/kg) High dose 450 mg (9 mg/kg) RR (95% CI) RD% (95% CI) Low dose 300 mg (6 mg/ kg) High dose 450 mg (9 mg/ kg) RR (95% CI) RD% (95% CI) Partici- pants with available data Outcomes 1–4: 127 Outcomes 5–10: 179 Outcomes 1–4: 124 Outcomes 5–10: 181 – – Varied by outcome Varied by outcome – – Treatment success 84/127 (66.14%) 69/124 (55.64%) 0.84 (0.69– 1.03) –10.50 (–22.50– 1.51) 84/101 83.17% 69/93 (74.19%) 0.89 (0.77– 1.03) –8.97 (–20.48– 2.53) Treatment failure 17/127 (13.39%) 24/124 (19.35%) 1.45 (0.82– 2.56) 5.97 (–3.16– 15.10) 17/101 (16.83%) 24/93 25.81%) 1.53 (0.88– 2.67) 8.97 (–2.53– 20.48) Relapse Not reported Not reported – – Not reported Not reported – – Death – total 4/127 (3.15%) 10/124 (8.06%) 2.56 (0.82– 7.95) 4.91 (–0.76– 10.60) 4/105 (3.81%) 10/98 (10.2%) 2.68 (0.87– 8.26) 6.39 (0.87– 1.01) Death – TB related 4/127 (3.15%) 8/124 (6.45%) 2.05 (0.63– 6.63) 3.30 (–1.98– 8.59) 4/105 (3.81%) 8/96 (8.33%) 2.19 (0.68– 7.03) 2.19 (–0.63– 13.42) Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review Annex 128 Adverse events – total 8/179 (4.47%) 12/181 (6.63%) 1.48 (0.62– 3.54) 2.16 (–2.56– 6.88) As per ITT analysis Serious adverse events 5/179 (2.79%) 5/181 (2.76%) 0.99 (0.29– 3.36) –0.03 (–3.43– 3.36) Adverse events requiring discontin- uation of treatment 4/179 (2.23%) 0/181 (0%) 1.76 (0.31– 10.19) –2.23 (–4.40– –0.07) Drug- induced liver Injury 2/179 (1.12%) 1/181 (0.55%) 0.49 (0.05– 5.40) –0.56 (–2.5– 1.32) Peripheral neuropa- thy 1/179 (0.56%) 0/181 (0%) 0.33 (0.01– 8.04) –0.56 (–1.65– 0.53) Adverse events – other 3/179 (1.68%) 7/181 (3.87%) 2.31 (0.61– 8.78) 2.19 (–1.19– 5.57) RR, relative risk; RD%, risk difference (%); 95% CI, 95% confidence interval Favez 1968 Favez et al. (124 participants) presented the results of a comparison of 5 mg/kg vs 10 mg/kg; all results of the trial for the ITT analysis are presented in this order. Higher doses may have resulted in little or no difference in treatment success (62.69% vs 59.65%, RR 0.95 [0.72–1.26], RD –3.04% [–20.25%–14.18%]) or treatment failure (1.49% vs 1.75%, RR 1.18 [0.08–18.37], RD 0.26% [–4.21%–4.74%]). The numbers of total deaths and TB-related deaths were not reported. The total number of adverse events may not have differed between the treatment groups (11.95% vs 12.28%, RR 1.03 [0.40–2.66], RD 0.34% [–11.19%–11.86%]), while the number of adverse events requiring treatment discontinuation might have been increased at the higher dose; however, the evidence for all safety outcomes was very uncertain (2.99% vs 5.26%, RR 1.76 [0.31–10.19], RD 2.28 [–4.81–9.36]). No drug-induced liver injury or peripheral neuropathy was reported in either group. The results of the PP analyses were consistent with those of the ITT analyses. See Table 3 and Annex 4 for full details. Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review Annex 1 29 Table 3. Findings from Favez 1968 Intention-to-treat analysis Per-protocol analysis Factor Low dose 5 mg/kg High dose 10 mg/ kg RR (95% CI) RD% (95% CI) Low dose 5 mg/kg High dose 10 mg/ kg RR (95% CI) RD (95% CI) Participants with available data 67 57 – – Varied by outcome Varied by outcome – – Treatment success 42/67 (62.69%) 34/57 (59.65%) 0.95 (0.72– 1.26 –3.04 (–20.25– 14.18) 42/43 (97.67%) 34/35 (97.14%) 0.99 (0.92– 1.07) –0.53 (–7.66– 6.59) Treatment failure 1/67 (1.49%) 1/57 (1.75%) 1.18 (0.08– 18.37) 0.26 (–4.21– 4.74) 1/43 (2.33%) 1/35 (2.86%) 1.23 (0.08– 18.94) 0.53 (–6.59– 7.66) Relapse Not reported Not reported – – Not reported Not reported – – Death – total Not reported Not reported – – Not reported Not reported – – Death – TB related Not reported Not reported – – Not reported Not reported – – Adverse events – total 8/67 (11.95%) 7/57 (12.28%) 1.03 (0.40– 2.66) 0.34 (–11.19– 11.86) As per ITT analysis Serious adverse events 8/67 (11.95%) 7/57 (12.28%) 1.03 (0.40– 2.66) 0.34 (–11.19– 11.86) Adverse events requiring dis- continuation of treatment 2/67 (2.99%) 3/57 (5.26%) 1.76 (0.31– 10.19) 2.28 (–4.81– 9.36) Drug- induced liver Injury 0/67 (0%) 0/57 (0%) 1.17 (0.02– 58.17) 0.00 (0.00– 0.00) Peripheral neuropathy 0/67 (0%) 0/57 (0%) 1.17 (0.02– 58.17) 0.00 (0.00– 0.00) Adverse events – other Not reported Not reported – – RR, relative risk; RD%, risk difference (%); 95% CI, 95% confidence interval Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review Annex 130 USPHS 1954 USPHS 1954 (373 participants) presented two comparisons of 3 mg/kg vs 10–12 mg/kg, each with a different background regimen. The pooled results are presented below. All the results for the ITT analysis are presented for the 3 mg/kg group followed by the 10–12 mg/kg group. Treatment success in the ITT analysis was probably lower at the higher dose (69.32% vs 57.36%, RR 0.83 [0.71–0.97], RD –11.96% [–21.66%– –2.26%]), but there was probably little or no difference in the PP analysis (84.72% vs 82.48%, RR 0.97 [0.88–1.07], RD –2.24 [–10.90–6.42]). There may have been little or no difference in treatment failure between the groups (12.5% vs 12.18%, RR 0.99 [0.58–1.70], RD –0.32 [–7.01–6.37]). As all-cause mortality was not reported, only TB-related deaths were available for analysis. This suggested little or no difference between the two dose groups (0.57% vs 1.02%, RR 1.42 [0.18–11.39], RD 0.45 [–1.34–2.23]), but the evidence was very uncertain. USPHS 1954 reported only adverse events that required discontinuation of treatment, which is assumed to represent total adverse events in the study. The rate of discontinuation may have been higher in the group at 10–12 mg/kg (3.41% vs 15.23%, RR 4.08 (1.56–10.64), RD 11.82 [6.13–17.51]), although the report did not describe specific adverse events of interest, including drug-induced liver injury and peripheral neuropathy. Except where noted, the results of the PP analyses were consistent with those of the ITT analyses, except for treatment success, as described above. See Table 4 and Annex 4 for full details. Table 4. Findings from USPHS 1954: pooled data from two comparison arms Intention-to-treat analysis Per-protocol analysis Factor Low dose 3 mg High dose 10–12 mg/kg RR (95% CI) RD% (95% CI) Low dose 3 mg High dose 10–12 mg/kg RR (95% CI) RD% (95% CI) Participants with available data 176 197 – – Varied by outcome Varied by outcome – – Treatment success 122/176 (69.32%) 113/197 (57.36%) 0.83 (0.71– 0.97 –11.96 (–21.66– –2.26) 122/144 (84.72%) 113/137 (82.48%) 0.97 (0.88– 1.07) –2.24 (–10.90– 6.42) Treatment failure 22/176 (12.50%) 24/197 (12.18%) 0.99 (0.58– 1.70) –0.32 (–7.01– 6.37) 22/144 (15.28%) 24/137 (17.52%) 1.15 (0.68– 1.95) 2.24 (–6.42– 10.90) Relapse Not reported Not reported – – Not reported Not reported – – Death – total Not reported Not reported – – Not reported Not reported – – Death – TB-related 1/176 (0.57%) 2/197 (1.02%) 1.42 (0.18– 11.39) 0.45 (–1.34– 2.23) 1/145 (0.69%) 2/139 (1.44%) 1.66 (0.21– 13.29) 0.75 (–1.65– 3.14) Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review Annex 1 31 Adverse Events – total 6/176 (3.41%) 30/197 (15.23%) 4.08 (1.56– 10.64) 11.82 (6.13– 17.51) As per ITT analysis Serious adverse events Not reported Not reported – – Adverse events requiring dis- continuation of treatment 6/176 (3.41%) 30/197 (15.23%) 4.08 (1.56– 10.64) 11.82 (6.13– 17.51) Drug- induced liver Injury Not reported Not reported – – Peripheral neuropathy Not reported Not reported – – Adverse events – other Not reported Not reported – – RR, relative risk; RD%, risk difference (%); 95% CI, 95% confidence interval Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review Annex 132 3. Discussion 3.1 Summary of main results This review addressed the question of whether higher doses of INH than those currently recommended in WHO guidance could safely improve outcomes for people receiving first- line treatment for TB. Four small trials (comprising a total of 901 participants) met the inclusion criteria for the review. The dose of INH ranged from 3 mg/kg to 10-12 mg/kg, but the dose comparisons evaluated in each trial were sufficiently different to preclude meaningful meta- analysis of any of the outcomes. The trial regimens also differed with respect to the other drugs included in the background regimen, the duration of treatment (10-18 months) and the times at which most outcomes were evaluated (5-12 months), limiting comparison with modern short- course regimens. The settings and study populations were heterogeneous. We therefore chose not to combine the data and to present the results of each trial separately. We found no evidence of a dose–response relation in the efficacy of doses of INH higher than those used in the current first-line treatment regimen. Although the treatment success rate probably improved when the dose of INH was increased from 4 to 6 mg/kg (EA/BMRC 1963), it was numerically or significantly lower in all the trials in which higher dose levels in the range of 9-12 mg/kg were evaluated (USPHS 1954, EA/BMRC 1966, Favez 1968). Similarly, with the exception of EA/BMRC 1963, treatment failure was probably similar at higher doses of INH. None of the trials reported on the relapse outcome. We found some evidence of increasing numbers of adverse events with higher doses of INH. Although, because of heterogeneity in the intervention, we were unable to combine the data, we found more adverse and/or serious adverse events at higher doses in all the studies, which was most marked in the studies of the highest doses (USPHS 1954 and Favez et al.). This was also true for treatment discontinuation in the studies that reported this outcome and was a statistically significant finding in USPHS 1954. Interpretation of these results is difficult because of the poor and inconsistent reporting of safety outcomes in these older trials; however, the differences were not attributed by the investigators to well-described dose-related side-effects of INH, such as drug-induced liver injury and peripheral neuropathy. The absence of a difference in efficacy may be valid, as dose-ranging studies of early bactericidal activity suggest that the dose–response relation for INH monotherapy plateaus at approximately 5 mg/kg per day. It could also, however, reflect underpowering due to the small size of each trial or numerous possible biases in these trials, particularly attrition bias due to relatively high losses to follow-up. The size of the trials and the poor quality of reporting could also contribute to the absence of differences in safety outcomes. Although well-described toxic effects of INH such as drug-induced liver injury and peripheral neuropathy are considered to be dose- and exposure-related, they are sufficiently infrequent that clinically important increases in the risk of their occurrence are unlikely to go undetected in smaller trials such as these. In addition, the lack of blinding and inconsistent ascertainment and reporting of safety outcomes in all the trials could have biased these safety assessments. Although there was no overall evidence of greater efficacy and some evidence of more adverse events in the trial populations as a whole, it is possible that some subgroups could benefit from the intervention. These are not well represented in the evidence presented in this review. The Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review Annex 1 33 metabolism of INH is highly polymorphic, with substantially different proportions of fast acetylators worldwide, for whom pharmacokinetics exposure is much lower, which could impact efficacy and safety. None of the trials described pharmacokinetics or pharmacogenetics that could identify or assess outcomes in this important subgroup, and it is notable that none enrolled participants in Asia, where the fast acetylator phenotype is commonest. Similarly, all the trials enrolled participants with pulmonary TB, and we found no relevant evidence of use of higher INH doses in serious forms of extrapulmonary disease such as meningeal TB, although one of the studies that was excluded explored administration of the currently recommended dose intravenously, which resulted in higher plasma pharmacokinetics and possibly lower mortality in people with both pulmonary and meningeal TB (44). This trial was, however, small, and higher doses of ethambutol were used simultaneously. All the trials were undertaken before the HIV pandemic, and HIV co-infected participants are not represented in this review. The trials also did not specifically report on the prevalence or impact of diabetes in their participants and did not exclude people with diabetes. 3.2 Overall completeness and applicability of evidence While all the studies reported treatment success, failure and at least one adverse event category, there were gaps in reporting some important outcomes. Deaths (total and/or TB-related) were either not reported or incompletely reported in three of the four studies (Favez 1968, EA/BMRC 1963 and USPHS 1954). Similarly, data on all adverse event outcomes, except those resulting in discontinuation and the total count, were missing from at least one study. In the case of USPHS 1954, this is probably due the fact that the publication appeared to be more of a “progress report” than a complete account of the trial. We did not identify a full report of the trial. Definitions of treatment success differed by study, and none reported relapses after treatment; we therefore cannot comment on the long-term efficacy of the interventions described. The trials included in this review were all completed before 1970, and none included RIF or PZA in the background regimen. It is therefore difficult to project how the results would compare with results obtained with the currently recommended first-line regimen. We interpreted the age of the trials as serious indirectness in applying the results to that with modern first-line treatment. Further, most treatment was delivered in hospitals (all in Favez 1968 and USPHS 1954, 6 months in EA/BMRC 1963 and 2 months in EA/BMRC 1966). This might also limit comparisons to many contemporary programmes of ambulatory care. Participants were tested for diabetes in only one study (EA/BMRC 1963), as an unofficial exclusion criterion. Given the well-known link between diabetes and TB infection, multi-drug-resistant TB, dying from TB and TB relapse (45, 46), the absence of diabetes and indeed of any description of comorbidities in participants in all the studies limits the applicability of our results. The trials did not report HIV status, as they precede the identification of AIDS (47). We excluded one study that involved different exposures to INH due to different routes of administration between arms (intravenous vs oral) (44); however, as the dose was the same in all arms, the study did not meet the inclusion criteria in our protocol. We also excluded another study of direct INH dose comparisons, as individualized dosing based on NAT2 genotyping was used (48). The authors claimed improved efficacy and reduced rates of hepatotoxicity with a dosing strategy of 2.5 mg/kg for slow, 5 mg/kg for intermediate and 7.5 mg/kg for fast acetylators as predicted by genotyping of the NAT*4 locus. We, however, decided to review trials of INH dose comparisons Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review Annex 134 without individualized or stratified therapy (with the exception of weight), as most high TB burden countries cannot routinely conduct therapeutic drug monitoring or genotyping. This trial therefore did not meet the inclusion criteria. 3.3 Certainty of the evidence All four trials were assessed as at high risk of bias overall, principally due to lack of blinding of investigators, participants and assessors and relatively high losses to follow-up. In addition, the background regimens used dated from before 1970 and differed from those currently used, limiting generalizability of the results to people on modern first-line regimens. All the trials were small, and we could not combine data on dose levels for any outcome, resulting in serious or very serious imprecision, as judged by optimal information size calculations, for all outcomes except treatment success in EA/BMRC 1966 and USPHS 1954. For these reasons, the certainty of the evidence was rated as low or very low for all other outcomes of interest. Because of the large number of possible comparisons and the impossibility of data synthesis, we did not create an overall summary of findings table. 3.4 Potential biases in the review process The scope of this review was narrow, focusing on changes in dosing in daily administration of drugs. We necessarily excluded many studies in which intermittent administration of higher doses of INH was used, which would be of historical interest to clinical trialists. Such dosing schedules were, however, usually calibrated to achieve a cumulative weekly mg/kg dose similar to daily regimens. As WHO now deprecates intermittent regimens and as the extent of aggregation of cumulative dosing is not known, use of these metrics could conceal important differences in pharmacokinetics and pharmacodynamics. We therefore did not examine that evidence in this review. We followed the standardized methods detailed in the Cochrane Handbook for Systematic Reviews (25). The literature search was conducted with a robust search strategy (detailed in Annex 1) in a variety of databases. We also searched the reference lists of the publications that were included and reviews on similar topics and used our own knowledge of the literature. All abstracts and full texts were screened for inclusion by two authors independently with the addition of a third author when required. We were unable to assess publication bias effectively because of the small number of studies included. We had planned to account for all patients lost to follow up by generating ITT and best- and worst-case analyses as appropriate. It was often reported that significant numbers of participants who were randomized and treated later defaulted but did not identify the treatment group to which they were assigned. Without this information, we were unable to conduct accurate sensitivity analyses to assess the impact of the missing data. The magnitude and direction of the effect that this might have on the findings cannot be estimated formally. Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review Annex 1 35 3.5 Agreements and disagreements with other studies and reviews As described above, we deliberately chose to review studies on INH dosing in which the intervention was as similar to the current first-line regimen as possible and the study populations in the different dosing arms were comparable. This was done so that the results would be as applicable as possible to the many low- and low–middle-income countries on the WHO list of countries with a high burden of TB (40) and that the results would be immediately actionable. To the authors’ knowledge, no similar review has been conducted. Other published work has addressed individualized or stratified strategies for dosing in general and for INH in particular. A recent systematic review of INH pharmacokinetics, for example, identified significantly lower INH AUC and Cmax values in TB patients than in healthy volunteers, an effect that was even stronger in those with HIV-TB co-infection (49). The evidence on whether use of therapeutic drug monitoring to alter INH dosing leads to better treatment outcomes remains, however, conflicting (50). A number of studies have been conducted on use of NAT2 genotyping in determining INH dosing. While these studies cannot be compared directly with our results, several have described comparable efficacy and toxicity with this approach (49). Routine NAT2 genotyping is not widely available, particularly in high-burden TB countries; however, Verma et al. (51) have reported a rapid assay for use with GeneXpert, which may be a first step to adoption of this novel strategy. 3.6 Authors’ conclusions Implications for practice We found conflicting evidence for whether higher doses of INH are associated with improved efficacy and some evidence that the rate of adverse events was increased; however, the overall level of certainty of the data was very low. This review therefore does not provide a rationale to change current WHO recommendations for dosing INH for people receiving first-line treatment for TB. We do not know whether higher doses are of benefit in serious forms of disease, such as meningeal or disseminated disease, nor whether other important subgroups, such as fast acetylators or HIV co-infected people, could benefit from the intervention. Implications for research Higher-quality, more directly applicable clinical trials of INH dosing in the context of the currently recommended first-line regimen in high-TB burden countries, with the currently recommended background regimen, would be desirable to inform future practice. Inclusion of a representative number of patients with HIV-TB co-infection and people with diabetes would be particularly important. The impact of the intervention on subgroups defined by acetylator status (from either pharmacokinetics and/or pharmacogenetics) would be useful to determine the possible value of stratified or individualized dosing strategies. Further research must focus not only on improving treatment success and preventing relapse but, equally importantly, on improving data on safety outcomes, including tolerability and treatment discontinuation. Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review Annex 136 3.7 Acknowledgements We acknowledge the WHO Global TB Programme, which commissioned and funded this review and provided valuable guidance and comments during its preparation. 3.8 References 1. Global tuberculosis report (WHO/CDS/TB/2019.15). Geneva: World Health Organization; 2019 (https:// www.who.int/tb/publications/global_report/en/). 2. Espinal MA, Kim SJ, Suarez PG, Kam KM, Khomenko AG et al. Standard short-course chemotherapy for drug-resistant tuberculosis: treatment outcomes in 6 countries JAMA. 2000;283:2537–45 (doi: 10.1001/jama.283.19.2537). 3. Fox W, Ellard GA, Mitchison DA. Studies on the treatment of tuberculosis undertaken by the British Medical Research Council tuberculosis units, 1946–1986, with relevant subsequent publications. Int J Tuberc Lung Dis. 1999;3(10 Suppl 2):S231–79. 4. British Thoracic and Tuberculosis Association. Short-course chemotherapy in pulmonary tuberculosis. A controlled trial by the British Thoracic and Tuberculosis Association. Lancet. 1976;2(7995):1102–4. 5. Lei B, Wei CJ, Tu SC. Action mechanism of antitubercular isoniazid. Activation by Mycobacterium tuberculosis KatG, isolation, and characterization of inha inhibitor. J Biol Chem. 2000;275:2520–6. 6. Ken-Dror G, Bonnet LJ, Koh G, Davies GR. Comparing the efficacy of first-line drug regimens for pulmonary tuberculosis: meta-analysis of phase III trials (in preparation 2022) 7. Guidelines for treatment of drug-susceptible tuberculosis and patient care (2017 update) (WHO/HTM/ TB/2017.05). Geneva: World Health Organization; 2017 (https://www.who.int/tb/publications/2017/ dstb_guidance_2017/en/). 8. Meeting report of a technical consultation on the pharmacokinetics and pharmacodynamics (PK/PD) of tuberculosis medicines WHO/HTM/TB/2017.28. Geneva: World Health Organization; 2017 (https:// apps.who.int/iris/handle/10665/260440 9. Ellard G, Gammon P. Pharmacokinetics of isoniazid metabolism in man. J Pharmacokinet Biopharm. 1976;4:83–113. 10. Richardson M, Kirkham J, Dwan K, Sloan DJ, Davies G, Jorgensen AL. NAT2 variants and toxicity related to anti-tuberculosis agents: a systematic review and meta-analysis. Int J Tuberc Lung Dis. 2019 23:293–305. 11. Krishnamurthy DV, Selkon JB, Ramachandran K, Devadata S, Mitchison DA, Radakrishna S et al. Effect of pyridoxine on vitamin B6 concentrations and glutamic-oxaloacetic transaminase activity in whole blood of tuberculous patients receiving high-dosage isoniazid. Bull World Health Organ. 1967;36:853–70. 12. Donald PR, Sirgel FA, Botha FJ, Seifart HI, Parkin DP, Vandenplas ML et al. The early bactericidal activity of isoniazid related to its dose size in pulmonary tuberculosis. Am J Respir Crit Care Med. 1997;156:895–900 (doi: 10.1164/ajrccm.156.3.9609132). 13. Donald PR, Sirgel FA, Venter A, Parkin DP, Seifart HI, van de Wal BW et al. The influence of human N-acetyltransferase genotype on the early bactericidal activity of isoniazid. Clin Infect Dis. 2004;39:1425– 30 (doi: 10.1086/424999). 14. Dooley KE, Miyahara S, von Groote-Bidlingmaier F, Sun X, Hafner R, Rosenkranz SL et al. Early bactericidal activity of different isoniazid doses for drug resistant TB (INHindsight): A randomized open- label clinical trial. Am J Respir Crit Care Med. 2020;201:1416–24. 15. Lienhardt C, Cook SV, Burgos M, Yorke-Edwards V, Rigouts L, Anyo G et al. Efficacy and safety of a 4-drug fixed-dose combination regimen compared with separate drugs for treatment of pulmonary tuberculosis: the Study C randomized controlled trial. JAMA. 2011;305:1415–23 (doi: 10.1001/ jama.2011.436). Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review Annex 1 37 16. Mitchison DA. Basic mechanisms of chemotherapy. Chest. 1979;76(S6):771–81. 17. Smieja M, Marchetti C, Cook D, Smaill FM. Isoniazid for preventing tuberculosis in non‐HIV infected persons. Cochrane Data Syst Rev. 1999; CD001363 (https://doi.org/10.1002/14651858). 18. Ken-Dror G, Davies GR, Bonnet LJ, Nevitt SJ. Exploring optimal first-line treatment for tuberculosis: a network meta-analysis (in preparation 2022) 19. Donald PR, Sirgel FA, Botha FJ, Seifart HI, Parkin DP, Vandenplas ML, Van de Wal BW, Maritz JS, Mitchison DA. The early bactericidal activity of isoniazid related to its dose size in pulmonary tuberculosis. Am J Respir Crit Care Med. 1997 ;156(3 Pt 1):895-900. 20. Donald PR, Sirgel FA, Venter A, Parkin DP, Seifart HI, van de Wal BW, Werely C, van Helden PD, Maritz JS. The influence of human N-acetyltransferase genotype on the early bactericidal activity of isoniazid. Clin Infect Dis. 2004;39(10):1425-30. 21. Gumbo T, Louie A, Liu W, Brown D, Ambrose PG, Bhavnani SM et al. Isoniazid bactericidal activity and resistance emergence: integrating pharmacodynamics and pharmacogenomics to predict efficacy in different ethnic populations. Antimicrob Agents Chemother. 2007;51:2329–36 22. Moher D, Liberati A, Tetzlaff J, Altman DG, The PRISMA Group Preferred Reporting Items for Systematic Reviews and Meta-Analyses: The PRISMA Statement. PLoS Med 2009; : e1000097. https://doi. org/10.1371/journal.pmed.1000097 23. Moher D, Shamseer L, Clarke M, Ghersi D, Liberati A, Petticrew M, Shekelle P, Stewart LA. Preferred Reporting Items for Systematic Review and Meta-Analysis Protocols (PRISMA-P) 2015 statement. Syst Rev. 2015; 4:1. doi: 10.1186/2046-4053-4- 24. Definitions and reporting framework for tuberculosis – 2013 revision (WHO/HTM/TB/2013.2). Geneva: World Health Organization; 2013 (https://apps.who.int/iris/bitstream/handle/10665/79199/9789241505345_ eng.pdf). 25. Lefebvre C, Glanville J, Briscoe S, Littlewood A, Marshall C, Metzendorf MI et al. Technical supplement to Chapter 4: Searching for and selecting studies. In: Higgins JPT, Thomas J, Chandler J, Cumpston MS, Li T, Page MJ et al., editors. Cochrane handbook for systematic reviews of interventions, version 6. Oxford: Cochrane; 2019. 26. Bonnett LJ, Ken-Dror G, Koh GCKW, Davies GR. Comparing the efficacy of drug regimens for pulmonary tuberculosis: Meta-analysis of endpoints in early-phase clinical trials. Clin Infect Dis. 2017;65:46–54 (doi: 10.1093/cid/cix247). 27. Covidence systematic review software 2020. Melbourne: Veritas Health Innovation (www.covidence. org). 28. Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. BMJ. 2003;327:557–60. 29. Jüni P, Altman DG, Egger M. Systematic reviews in health care: assessing the quality of controlled clinical trials. BMJ. 2001;323(7303):42–6. 30. Deeks JJ, Higgins JP, Altman DG, editors, on behalf of the Cochrane Statistical Methods Group. Chapter 9: Analysing data and undertaking meta-analyses. In: Higgins JPT, Churchill R, Chandler J, Cumpston MS, editors. Cochrane handbook for systematic reviews of interventions, version 5.2.0 (updated June 2017). Oxford: Cochrane; 2019 (www.training.cochrane.org/handbook). 31. Higgins JP, Altman DG, Gøtzsche PC, Jüni P, Moher D, Oxman AD et al. The Cochrane Collaboration’s tool for assessing risk of bias in randomised trials. BMJ. 2011;343:d5928 (doi: 10.1136/bmj.d5928). 32. DerSimonian R, Laird N. Meta-analysis in clinical trials. Controlled Clin Trials. 1986;7:177–88. 33. R Core Team. R: A language and environment for statistical computing. Vienna: R Foundation for Statistical Computing; 2018 (https://www.R-project.org/). 34. Guyatt G, Oxman AD, Vist G, Kunz R, Falck-Ytter Y, Alonso-Coelle P et al. GRADE: an emerging consensus on rating quality of evidence and strength of recommendations. BMJ. 2008;336:3. 35. GRADEpro GDT: GRADEpro Guideline Development Tool [Software]. Hamilton (ON): McMaster University; 2020 (gradepro.org). Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review Annex 138 36. East African and British Medical Research Councils. Isoniazid with thiacetazone in the treatment of pulmonary tuberculosis in East Africa – second investigation. Tubercle. 1963;44:301–33. 37. East African and British Medical Research Councils. Isoniazid with thiacetazone (thioacetazone) in the treatment of pulmonary tuberculosis in East Africa – fourth investigation. Tubercle. 1966;47:315–39. 38. Favez Z, Maillard JM, Vouilloz M. Résultats comparés de doses moyennes et élevées d’isoniazide dans le traitement triple de la tuberculose pulmonaire ouverte [Comparative results of medium and high doses of isoniazid in triple treatment of open pulmonary tuberculosis]. Schweiz Med Wochensch. 1968;98(36):1392–5. 39. United States Public Health Service Cooperative. Progress report on therapeutic and toxic effects of combination of isoniazid, streptomycin and para-aminosalicylic acid. Am Rev Tuberc. 1954;69(1):1–12. 40. WHO releases new global lists of high-burden countries for TB, HIV-associated TB and drug-resistant TB. Geneva: World Health Organization; 2021 (https://www.who.int/news/item/17-06-2021-who- releases-new-global-lists-of-high-burden-countries-for-tb-hiv-associated-tb-and-drug-resistant-tb). 41. World Bank country and lending groups. Washington DC: World Bank; 2021 (https://datahelpdesk. worldbank.org/knowledgebase/articles/906519-world-bank-country-and-lending-groups). 42. WHO consolidated guidelines on tuberculosis, module 4: Treatment – drug-resistant tuberculosis treatment. Geneva: World Health Organization; 2020 43. Iseman MD. Tuberculosis therapy: past, present and future. Eur Respir J. 2002;36(Suppl):87s–94s (doi: 10.1183/09031936.02.00309102). 44. Butov D, Feshchenko Y, Kuzhko M, Gumenuik, M, Yurko, K, Grygorova, A et al Effectiveness of intravenous isoniazid and ethambutol administration in patients with tuberculosis meningoencephalitis and HIV infectionTuberculosis and respiratory diseases 2019; 82:1-8 45. Baker MA, Harries AD, Jeon CY, Hart JE, Kapur A, Lönnroth K et al. The impact of diabetes on tuberculosis treatment outcomes: a systematic review. BMC Med. 2011;9:81 (doi: 10.1186/1741-7015-9-81). 46. Tegegne BS, Mengesha MM, Teferra AA, Awoke MA, Habtewold TD. Association between diabetes mellitus and multi-drug-resistant tuberculosis: evidence from a systematic review and meta-analysis. Syst Rev. 2018;7(1):161 (doi: 10.1186/s13643-018-0828-0). 47. Sharp PM, Hahn BH. Origins of HIV and the AIDS pandemic. Cold Spring Harb Perspect Med. 2011;1(1):a006841 (doi: 10.1101/cshperspect.a006841). 48. Azuma J, Ohno M, Kubota R, Yokota S, Nagai T, Tsuyuguchi K et al. NAT2 genotype guided regimen reduces isoniazid-induced liver injury and early treatment failure in the 6-month four-drug standard treatment of tuberculosis: a randomized controlled trial for pharmacogenetics-based therapy. Eur J Clin Pharmacol. 2013;69(5):1091–101 (doi: 10.1007/s00228-012-1429-9). 49. Hong BL, D’Cunha R, Li P, Al-Shaer MH, Alghamdi WA, Guohua A et al. A systematic review and meta-analysis of isoniazid pharmacokinetics in healthy volunteers and patients with tuberculosis. Clin Ther. 2020;42(11):e220–42 (doi: 10.1016/j.clinthera.2020.09.009). 50. Verbeek RK, Günther G, Kibuule D, Hunter C, Rennie TW. Optimizing treatment outcome of first-line anti-tuberculosis drugs: the role of therapeutic drug monitoring Eur J Clin Pharmacol. 2016;72(8):905– 16 (doi: 10.1007/s00228-016-2083-4). 51. Verma R, Patil S, Zhang N, Moreira FMF, Vitorio MT, Santos AS et al. A rapid pharmacogenomic assay to detect NAT2 polymorphisms and guide isoniazid dosing for tuberculosis treatment (awaiting peer review 2021) Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review Annex 1 39 Appendix 1. Search strategy Search Set Central Medline Embase 1 Tuberculosis [tak] Tuberculosis [ms] Tuberculosis [mt] 2 Rifampicin [tak] OR Isoniazid [tak] OR Pyrazinamide [tak] OR Ethambutol [tak] Tuberculosis [tiab] Tuberculosis [tiab] 3 1 AND 2 1 OR 2 1 OR 2 4 Latent [tak] OR Prevention [tak] Rifampicin [tiab] OR Rifampin [tiab] OR Isoniazid [tiab] OR Pyrazinamide [tiab] OR Ethambutol [tiab] Rifampicin [tiab] OR Rifampin [tiab] OR Isoniazid [tiab] OR Pyrazinamide [tiab] OR Ethambutol [tiab] 5 3 NOT 4 Rifampin [sc] OR Isoniazid [sc] OR Pyrazinamide [sc] OR Ethambutol [sc] Rifampicin [mt] OR Isoniazid [mt] OR Pyrazinamide [mt] OR Ethambutol [mt] 6 4 OR 5 4 OR 5 7 3 AND 6 3 AND 6 8 Randomized controlled trial [pt] OR Controlled clinical trial [pt] Crossover procedure [de] OR Double-blind procedure [de] OR Randomized controlled trial [de] OR Single-blind procedure [de] 9 Clinical trials as topic [mesh: no exp] Random* [deabti] OR Factorial* [deabti] OR Crossover* [deabti] OR Cross NEXT/1 over* [deabti] OR Placebo* [deabti] OR Doubl* NEAR/1 blind* [deabti] OR Singl* NEAR/1 blind* [deabti] OR Assign* [deabti] OR Allocat* [deabti] OR Volunteer* [deabti] 10 Randomized [tiab] OR Randomly [tiab] OR placebo [tiab] OR trial [ti] 8 OR 9 11 8 OR 9 OR 10 Animal experiment [mt] NOT (Human experiment [mt] OR Human [mt]) 12 Animals [mh] NOT Humans [mh] 10 NOT 11 13 11 NOT 12 7 AND 12 Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review Annex 140 Search Set Central Medline Embase 14 7 AND 13 Latent [tiab] OR Prevention [tiab] 15 Latent [tiab] OR Prevention [tiab] 13 NOT 14 16 14 NOT 15 [tak] word in title abstract or keyword, [pt] Publication Type, [ti ] word in title, [tiab] word in title or abstract, [sh] subheading, [mh] exploded MeSH term, [mesh: noexp] unexploded MeSH term, [mt] exploded Emtree term, [de] design, [deabti] Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review Annex 1 41 Appendix 2. Characteristics of the studies included EA/BMRC 1963 Methods Randomized controlled trial of high- and low-dose INH in combination with high or low dose thiacetazone for the treatment of pulmonary TB. One control arm of standard dose INH in combination with PAS. Patients were eligible for change of chemotherapy if sputum positive at 6 months. Follow-up: Clinical observations at monthly intervals. Sputum samples sent for testing at baseline then at month 3, 6, 9, 10, 11 and 12. Lost to follow: At the end of 12 months of treatment, 36 had been lost to follow-up and 2 withdrew. Participants Setting: All patients treated as inpatients for the first 6 months then as outpatients for the second 6 months. All treated in either Kenya, Uganda or the United Republic of Tanzania. No further information on the inpatient facilities reported. Number of participants: 334 randomized of whom 34 had INH-resistant disease and are reported elsewhere. Number in each arm therefore as follows: – 10PH: 75 – T150/H300: 73 – T150/H200: 80 – T100/H300: 72 Note – two groups extracted for review inclusion were T150/300 and T150/200 Inclusion criteria: – Acute extensive pulmonary TB of recent onset – Positive sputum for TB on direct smear examination – Age ≥ 15 years – “Considered to be co-operative and accessibly domiciled” – Not previously received anti-TB chemotherapy Exclusion criteria: – Weight < 36.3 kg – Very poor general condition – Haemoglobin < 6 g/100 mL – Extrapulmonary TB requiring treatment – Pleural effusion obscuring more than a third of the lung HIV status: No testing reported Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review Annex 142 Baseline drug susceptibility (data reported only for participants who completed 6 months of their assigned treatment (n=278)): – INH: 34 resistant, results reported in another publication – Thiacetazone: 10 resistant – para-Aminosalicylic acid: 5 resistant Age and gender (data reported only for participants who completed 6 months of their assigned treatment (n=278)): M:F 203:75, Median age (category) = 25–34 years Interventions Group 1: INH 200 mg (4 mg/kg) daily (OD) and PAS 5 g twice daily (BD) Group 2: INH 300 mg (7 mg/kg) OD and thiacetazone 150 mg OD Group 3: INH 200 mg (4 mg/kg) OD and thiacetazone 150 mg OD Group 4: INH 300 mg (7 mg/kg) and thiacetazone 100 mg OD All drugs given for 12 months Other interventions: none reported Outcomes Outcomes of interest: Deaths: overall and TB related Bacteriological results at 0, 3 and 6 months Radiographic changes Drug resistance patterns Urine testing for INH Weight changes Favourable status (treatment success) at 12 months defined as any participant with: – 3 single negative sputum cultures at any of the following times: • 10, 11 and 12 months • 9, 10 and either 11 or 12 months • 6, 9 and either 10, 11 or 12 months • 3, 5 and either 9, 10, 11 or 12 months – 2 single negative sputum cultures at 2 of the following months – 9, 10, 11 or 12. Unfavourable status (treatment failure) defined as: – Patients with 1 positive sputum culture among those at 9, 10, 11 or 12 months – Patient whose chemotherapy was changed because of bacterial positivity or clinical deterioration – Patients who died from TB – Note – only the former two will be used to define treatment failure in the full text of the review. Paper also included “patients with doubtful status”, from which we did not extract, as the data are not assignable to treatment success or failure. Adverse events: including peripheral neuropathy, drug-related liver injury (defined as “jaundice” rather than biochemically), serious adverse events, those requiring discontinuation of treatment and overall numbers. All presented as number of people experiencing the adverse event. Other reported outcomes: Bacteriological results at 0, 3 and 6 months Radiographic changes Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review Annex 1 43 Drug resistance patterns Urine testing for INH Weight changes Risk of bias Bias Authors’ judgement Support for judgement Random sequence generation (selection bias) Low risk Clearly described adequate method of randomization: pre-arranged series of sealed envelopes based on random sampling numbers Allocation concealment (selection bias) Low risk As above Blinding of participants and personnel (performance bias) All outcomes High risk Insufficient reporting, no mention of blinding measures Blinding of outcome assessment (detection bias) All outcomes High risk Only radiologist blinded Incomplete outcome data (attrition bias) All outcomes High risk 17 participants lost to follow-up and 2 withdrew in first 6 months, all with no reasons described. All of these participants were included in the analysis. 19 participants lost to follow-up in second 6 months of study, but their groups were not identified, so these were not included in the analysis. Selective reporting (reporting bias) Unclear Most outcomes reported. Non-TB deaths not reported for months 6-12 of the study. Other bias Of the 300 participants presented who had fully susceptible TB, 228 stayed on their allocated treatment for the full 12 months. The authors, however, presented results at 12 months for all 300 by “attempt[ing] to estimate what the response of the patients would have been at 12 months if they had continue[d] on their prescribed regiment throughout the trial”. The way in which these estimates were calculated is not reported. Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review Annex 144 EA/BMRC 1966 Methods Randomized controlled trial with eight active arms comprising combinations of the following variables in treating pulmonary TB: – High or low dose INH – Inpatient treatment for initial 2 months or outpatient treatment throughout – Two surprise home visits per month or no such visits Follow-up: Sputum samples sent for culture at monthly intervals, clinical observation frequency and details not provided. Lost to follow-up: 91 lost to follow-up. Participants Setting: Patients recruited in Kenya, Uganda and United Republic of Tanzania. 125 patients treated as inpatients for initial 2 months, 126 treated as outpatients throughout. No further detail on inpatient environment reported. Number of participants: Outcomes 1-4 (treatment success, treatment failure, total deaths, TB-related deaths) Outcome results 1-4 for 122 patients not reported for the following reasons: – 47 for “pre-treatment factors”, e.g., INH resistance – 71 who defaulted in the first 6 months – 2 with no data for the outcome – 2 who changed chemotherapy (one by accident and another on patient insistence not related to toxicity) These 122 patients are not represented in our analysis of outcomes 1-4 as the group allocations of those with H resistance and those who were ineligible for inclusions according to our protocol (e.g., having already started TB treatment) cannot be excluded. Furthermore, the outcomes of some of the 75 other patients were available to the authors but were not presented in the publication; Thus, inclusion of these patients in an ITT analysis would be highly misleading. This left 251 patients (127 TH300 and 124 TH450) for the analysis of outcomes 1-4. Outcomes 5–10 (Adverse event outcomes) Data on adverse events are presented for all patients in the study (except for 13). While we cannot exclude those with H resistance and those who did not meet our inclusion criteria for the review, we included the data, as no alternative was possible. This leaves 373 patients (179 TH300 and 181 TH450) in the analysis for outcomes 5-10. Inclusion criteria: – Pulmonary TB of recent onset – Positive sputum for TB on direct smear examination – Age ≥ 15 years – “Considered to be co-operative and accessibly domiciled” – Not previously received anti-TB therapy Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review Annex 1 45 Exclusion criteria: – Weight < 36.3 kg – Very poor general condition – Haemoglobin < 6 g/100 mL – Extrapulmonary TB requiring treatment – Pleural effusion obscuring more than a third of the lung HIV status: No testing reported Baseline drug susceptibility (data reported only for participants who completed 6 months of their assigned treatment (n=251)): – INH resistance: n = 20 (all excluded from the report) – Thiacezatone resistance: not fully reported Age and gender (data reported only for participants who completed 6 months of their assigned treatment (n=251)): M:F 171:80; median age (category) < 35 years Interventions Group 1: INH 300 mg (7 mg/kg) OD and 150 mg thiacetazone OD Group 2: INH 450 mg (10 mg/kg) OD, 150 mg thiacetazone OD and pyridoxine 6 mg OD All drugs given for 18 months Outcomes Outcomes of interest: Favourable response (treatment success) defined as: – All cultures negative at 10, 11 and 12 months – Only one positive culture obtained at 10, 11 or 12 months Unfavourable response (treatment failure) defined as: – two or more positive cultures at 10, 11 or 12 months – Chemotherapy changed because of clinical deterioration in the presence of positive sputum culture – Chemotherapy changed for drug toxicity – Death from TB, with active TB or from drug toxicity – Note: only the former two were used to define treatment failure in the full text of the review. Death: all causes and TB related Adverse events: total number, serious adverse events, those requiring discontinuation of treatment, peripheral neuropathy, drug-induced liver injury (defined as “jaundice” rather than biochemically) and other. – Publishing author definition of “serious” is used, as the data were not presented in a way that allowed the authors of this review to judge the seriousness of the adverse events described. The exception is drug-induced liver injury, which is distinguishable from the serious adverse event category and is presented separately in our review. Other reported outcomes: Radiographic changes Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review Annex 146 Bacteriological changes at 0, 1, 2, 3, 6, 9 and 12 months Drug sensitivity testing Risk of bias Bias Authors’ judgement Support for judgement Random sequence generation (selection bias) Low risk Clearly described, adequate method of randomization: pre-arranged series of sealed envelopes based on random sampling numbers Allocation concealment (selection bias) Low risk As above Blinding of participants and personnel (performance bias) All outcomes High risk Insufficient reporting, no mention of blinding measures Blinding of outcome assessment (detection bias) All outcomes High risk Only X-ray interpreters blinded Incomplete outcome data (attrition bias) All outcomes High risk Randomized group of 75 patients not described, impairing analysis High patient default rate Selective reporting (reporting bias) Low risk All expected outcomes reported Other bias Nil Favez et al. 1968 Methods Randomized controlled trial of 5 mg/kg versus 10 mg/kg in treatment of pulmonary TB Follow-up: sputum cultures obtained monthly up to 5 months, and radiographs obtained monthly up to 7 months. Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review Annex 1 47 Lost to follow-up: No denominators in tables of results at follow-up, so times unknown. Participants Setting: Hospital based care in Lausanne, Switzerland Number of participants: 124 randomized, 67 to low-dose INH and 57 to high-dose INH Inclusion/exclusion criteria: Patients all admitted after December 1965, no other inclusion/ exclusion criteria reported HIV status: not reported Baseline drug susceptibilitiy: – INH: 9, excluded from trial – Streptomycin: 1, excluded from trial Age and gender: M:F= 80:44; median age (category). 31–40 years Interventions Group 1: INH 5 mg/kg Group 2: INH 10 mg/kg Both groups treated with pyridoxine 40 mg OD and the same backbone chemotherapy, as follows: – Streptomycin 1 g OD for 8 weeks, then 3 g once a week – Thiocarlide 6 g OD Duration of treatment not reported Other interventions: none reported Outcomes Outcomes of interest: Treatment success: defined as all cultures negative at 4 months Adverse events: total number (for streptomycin ototoxicity and INH peripheral neuropathy only) and number discontinuing treatment. All presented as numbers of people experiencing the adverse event. Risk of bias Bias Authors’ judgement Support for judgement Random sequence generation (selection bias) Unclear Exact method not described: “distribution au hasard … a un sujet sur deux” [random distribution … one subject out of two] Allocation concealment (selection bias) Unclear No description of methods for concealing allocation Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review Annex 148 Blinding of participants and personnel (performance bias) All outcomes High risk Neither participants nor investigators blinded Blinding of outcome assessment (detection bias) All outcomes High risk No explicit mention of blinding of laboratory or radiology staff Incomplete outcome data (attrition bias) All outcomes High risk Tables do not report on all 109 participants for any outcome, and there is no explicit mention of loss to follow-up or death. Selective reporting (reporting bias) High risk Methods do not mention prioritization of efficacy or safety end-points, and safety end-points are incompletely reported Other bias USPHS 1954 Methods Randomized controlled trial of high- versus low-dose INH in combination with either streptomycin, para-aminosalicylic acid or both, in the treatment of pulmonary TB Follow-up: At 4-weekly intervals, the following were performed: clinical observations, chest radiography, sputum culture and microscopy. Lost to follow-up: At outcome assessment, 72 patients had been lost to follow-up and 2 had withdrawn consent. Participants Setting: Patients treated at 22 participating hospitals in the USA, all as inpatients throughout treatment. No further information reported. Number of participants: 461 randomized and eligible for outcome measurement at 20 weeks, distributed as follows: – 3H/S: 91 – 3H/PAS: 85 – 10H/S: 95 – 10H/PAS: 102 – 10H/S/PAS: 86 Only the first for groups were included for analysis in this review. Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review Annex 1 49 Inclusion criteria: – Pulmonary TB – No previous treatment Exclusion criteria: not reported HIV status: not reported Baseline drug susceptibility: not reported Age and gender (data reported only for participants who completed 20 weeks of their assigned treatment (n=335)): M:F 204:131. Median age (category), 33–54 years Interventions Group 1 (3H/S): INH 3 mg/kg OD and streptomycin 1 g twice a week Group 2 (3H/PAS): INH 3 mg/kg OD and PAS 10-12 g OD Group 3 (10H/S): INH 10 mg/kg OD and streptomycin 1 g twice a week Group 4 (10H/PAS): INH 3 mg/kg OD and PAS 10-12 g OD Group 5 (10H/S/PAS): INH 3 mg/kg OD, streptomycin 1 g twice a week and PAS 10-12 g OD All drugs given for 40 weeks Outcomes Outcomes of interest: Treatment success: defined as culture-negative status at 16 weeks Treatment failure: defined as culture-positive status at 16 weeks, including those who had treatment changed for clinical deterioration Deaths from TB Adverse events: total number and those requiring discontinuation of treatment. All presented as number of people experiencing the adverse event. – The paper does report some patients who experienced peripheral neuropathy; however, the data are presented unclearly, and reliable extraction was not possible. This outcome is therefore recorded as not reported. Other reported outcomes: Course of fever Body weight Chest X-ray changes Risk of bias Bias Authors’ judgement Support for judgement Random sequence generation (selection bias) Unclear Insufficient reporting Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review Annex 150 Allocation concealment (selection bias) Unclear Insufficient reporting Blinding of participants and personnel (performance bias) All outcomes High risk Insufficient reporting, no mention of blinding measures Blinding of outcome assessment (detection bias) All outcomes High risk Insufficient reporting, no mention of blinding measures Incomplete outcome data (attrition bias) All outcomes High risk Large volume of patients lost to follow-up 11 patients unaccounted for due to lack of reporting Selective reporting (reporting bias) High risks Deaths not related to TB not reported Sputum culture results at 20 weeks not reported Other bias Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review Annex 1 51 Appendix 3. Studies that were excluded Study Reason for exclusion Azuma 2013 (1) Wrong population Butov 2020 (2) Wrong intervention Kuzhko 2017 (3) Wrong intervention Kuzhko 2016 (4) Wrong intervention Kraan 1969 (5) Wrong population Ohno 2013 (6) Duplicate TCC 1973 (7) Wrong intervention TRC 1981 (8) Wrong intervention Wu 2011 (9) Report very unclear, data not extractable References 1. Azuma J, Ohno M, Kubota R, Yokota S, Nagai T, Tsuyuguchi K, Okuda Y, Takashima T, Kamimura S, Fujio Y, Kawase I; Pharmacogenetics-based tuberculosis therapy research group.NAT2 genotype guided regimen reduces isoniazid-induced liver injury and early treatment failure in the 6-month four- drug standard treatment of tuberculosis: a randomized controlled trial for pharmacogenetics-based therapy. Eur J Clin Pharmacol. 2013 ;69(5):1091-101 2. Butov D, Feshchenko Y, Kuzhko M, Gumenuik, M, Yurko, K, Grygorova, A et al Effectiveness of intravenous isoniazid and ethambutol administration in patients with tuberculosis meningoencephalitis and HIV infection Tuberculosis and Respiratory Diseases 2019; 82:1-8 3. Kuzhko M, Gumeniuk M, Butov D, Tlustova T, Denysov O, Sprynsian T. Features of intravenous anti TB therapy in patients with first diagnosed pulmonary TB in the intensive phase of treatment. European respiratory journal 2017;50 (Suppl 61) European Respiratory Society Annual Congress 2017 Abstract PA3496 4. Kuzhko M, Hulchuk N, Tlustova T, Avramchuk O, Gumeniuk M. The effectiveness of pulmonary TB treatment in patients with organic liver diseases, depending on the way of administration of anti-TB drugs. European respiratory journal 2016 ; 48 (Suppl 60) European Respiratory Society Annual Congress 2016 Abstract PA2671 5. Kraan JK, Mulder RJ, van Dijk B. Controlled study on rifampicin in first treatment of fresh cases of pulmonary tuberculosis Acta Tuber Pneumol Belgica 1969 3-4: 557-562 6. Ohno, M.; Azuma, J.; Yokota, S. Therapeutic potential of the NAT2 genotype-guided dosing stratification of isoniazid in chemotherapy for tuberculosis. Therapeutic Drug Monitoring 2013;35 (5):710 7. Tuberculosis Chemotherapy Centre, Madras. Controlled comparison of oral twice-weekly and oral daily isoniazid plus PAS in newly diagnosed pulmonary tuberculosis. British medical journal 1973;2(5857):7-11 8. Tuberculosis Chemotherapy Centre, Madras. Ethambutol plus isoniazid for the treatment of pulmonary tuberculosis--a controlled trial of our regimens. Tubercle 1981;62(1):13-29 9. Wu, F. High doses isoniazid and extend to strengthen the period of tuberculous meningitis effectiveness studies. Respirology 2011;2:82 Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review Annex 152 Appendix 4. Data and analyses Analysis 1. Treatment success, intention-to-treat analysis of relative risk of low dose vs high dose INH Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review Annex 1 53 Analysis 2. Treatment success, per protocol analysis of relative risk of low dose vs high dose INH Analysis 3. Treatment failure, intention-to-treat analysis of relative risk of low dose vs high dose INH Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review Annex 154 Analysis 4. Treatment failure, per-protocol analysis of relative risk of low dose vs high dose INH Analysis 5. All-cause deaths, intention-to-treat analysis of relative risk of low dose vs high dose INH Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review Annex 1 55 Analysis 6. All causes of death, per-protocol analysis of relative risk of low dose vs high dose INH Analysis 7. TB-related death, intention-to-treat analysis of relative risk of low dose vs high dose INH Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review Annex 156 Analysis 8. TB-related deaths, per-protocol analysis of relative risk of low dose vs high dose INH Analysis 9. Total adverse events, intention-to-treat analysis of relative risk of low dose vs high dose INH Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review Annex 1 57 Analysis 10. Serious adverse events, intention-to-treat analysis of relative risk of low dose vs high dose INH Analysis 11. Adverse events requiring drug discontinuation, intention-to-treat analysis of relative risk of low dose vs high dose INH Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review Annex 158 Analysis 12. Drug-induced liver injury, intention-to-treat analysis of relative risk of low dose vs high dose INH Analysis 13. Drug-induced peripheral neuropathy, intention-to-treat analysis of relative risk of low dose vs high dose INH Optimization of the dose of isoniazid in adults with presumed drug-susceptible tuberculosis: systematic review Annex 1 59 Analysis 14. Other adverse events, intention-to-treat analysis of relative risk of low dose vs high dose INH

Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis ANNEX 2 K. Haigh1, H. Twabi2, H. Ryan1, V. Lutje3, S. Nevitt1, G. Davies1 1 University of Liverpool, United Kingdom 2 College of Medicine, University of Malawi, Malawi 3 Liverpool School of Tropical Medicine, United Kingdom

Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2 63 Contents 1. Background 64 1.1 Description of the condition 64 1.2 Description of the intervention 64 1.3 How the intervention might work 65 1.4 Why this review is important 66 1.5 Methods 66 2. Results 72 2.1 Description of studies 72 2.2 Effects of interventions 78 3. Discussion 82 3.1 Summary of main results 82 3.2 Overall completeness and applicability of evidence 83 3.3 Certainty of the evidence 83 3.4 Limitations 84 3.5 Potential biases in the review process 84 3.6 Agreements and disagreements with other studies or reviews 84 3.7 Authors’ conclusions 85 3.8 Acknowledgements 86 3.9 References 86 Appendix 1. Search strategy 89 Appendix 2. Characteristics of the studies included 91 Appendix 3. Studies that were excluded from the review 119 Appendix 4. Funnel and Galbraith plots 121 Treatment success 121 All-cause mortality 121 Appendix 5. Data and analyses 123 Analysis 1. Treatment success: intention-to-treat analysis of risk ratio for rifampicin at doses of 10 mg/kg versus 15 mg/kg 123 Analysis 2. Treatment success: intention-to-treat analysis of risk ratio for rifampicin at doses of 10 mg/kg versus 20 mg/kg 123 Analysis 3. Treatment success: per-protocol analysis of risk ratio for rifampicin at doses of 10 mg/kg versus 15 mg/kg and 10 mg/kg versus 20 mg/kg 124 Analysis 4. Treatment failure: intention-to-treat analysis of risk ratio for rifampicin at doses of 10 mg/kg versus 15 mg/kg and 10 mg/kg versus 20 mg/kg 124 Analysis 5. Relapse: intention-to-treat analysis of risk ratio for rifampicin at doses of 10 mg/kg versus 15 mg/kg and 10 mg/kg versus 20 mg/kg 124 Analysis 6. All-cause mortality: intention-to-treat analysis of risk ratio for rifampicin at doses of 10 mg/kg versus 15 mg/kg 125 Analysis 7. All-cause mortality: intention-to-treat analysis of risk ratio for rifampicin at doses of 10 mg/kg versus 20 mg/kg 125 Analysis 8. All-cause mortality: intention-to-treat analysis of risk ratio for rifampicin at doses of 10 mg/kg versus 30 mg/kg and 10 mg/kg versus 35 mg/kg 125 Analysis 9. All-cause mortality: per-protocol analysis of risk ratio of rifampicin at doses of 10 mg/kg versus 15 mg/kg and 10 mg/kg versus 20 mg/kg 126 Analysis 10. Serious adverse events: intention-to-treat analysis of risk ratio for rifampicin at doses of 10 mg/kg versus 15 mg/kg and 10 mg/kg versus 20 mg/kg 126 Analysis 11. Drug-induced liver injury: intention-to-treat analysis of risk ratio for rifampicin at doses of 10 mg/kg versus 15 mg/kg 127 Analysis 12. Drug-induced liver injury: intention-to-treat analysis of risk ratio for rifampicin at doses of 10 mg/kg versus 20 mg/kg 127 Analysis 13. Drug-induced liver injury: intention-to-treat analysis of risk ratio for rifampicin at doses of 10 mg/kg versus 30 mg/kg and 10 mg/kg versus 35 mg/kg 128 Appendix 6. Summary of findings 129 Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 264 1. Background 1.1 Description of the condition Tuberculosis (TB) remains the single biggest killer of all infectious diseases, with an estimated 10 million cases and 1.2 million deaths worldwide in 2018 (1). While 30 low- and middle-income countries account for almost 90% of this burden, a significant fraction can be attributed to risk factors such as HIV co-infection (9%) and diabetes (15%). While progress has recently been made in the control of TB, the annual decrease in incidence does not currently exceed 2%. Seven million people were treated with first-line TB regimens globally in 2018, with a treatment success rate of 85%, but estimates suggest that the case fatality rate among all incident cases was as high as 15%, with most of the deaths among people in the most economically active period of their lives (1). Although pulmonary disease is the commonest presentation of TB, extrapulmonary and disseminated disease, particularly with neurological involvement, may be associated with worse outcomes. In addition, some people with TB harbour resistant strains of Mycobacterium tuberculosis for which first-line therapy is not effective (2). Of those presenting with TB for the first time, 13.1% are resistant to isoniazid (INH) and 3.3% are resistant to rifampicin (RIF), requiring modified, longer, more expensive treatment regimens (1). 1.2 Description of the intervention First-line short-course regimens for TB have changed little since their introduction more than 40 years ago. The finding that when RIF and pyrazinamide (PZA) were added to INH the duration of therapy could be reduced to as little as 6 months established this trio of drugs as the backbone of effective therapy (3). A fourth drug was often used to prevent the emergence of resistance, particularly in people who already harboured INH-resistant strains. While streptomycin was initially widely used for this purpose, widespread resistance and the need for parenteral administration led to its replacement by the oral agent ethambutol (ETH), resulting in the current standard first- line regimen (4). Rifampicin (RIF) is a semi-synthetic derivative of rifamycin SV, a natural product of Amycolatopsis mediterranei, discovered in 1967. It was widely incorporated into TB regimens in the late 1970s following the results of pivotal clinical trials and remains the most important drug in first-line TB regimens. RIF-based regimens reduce the length of treatment from 18 to 9 and 6 months in the absence and presence of PZA, respectively (3), and the duration of RIF administration in a regimen is a critical determinant of relapse after treatment (5). Resistance to RIF, of the four component drugs, is independently associated with the highest risk of poor outcome after first- line treatment (RR 5.5) (2). RIF has usually been used at a dose of 10 (8–12) mg/kg, adjusted for weight according to banding schedules depending on whether it is formulated as a single drug or as part of a fixed- dose combination (6). Similar doses have been used in both daily and intermittent treatment regimens. The rationale for the choice of this dose appears to have been based on pragmatic considerations of risk–benefit and cost. Perceptions of risk of immune-mediated side-effects (when used in intermittent regimens) and of drug-induced liver injury at higher doses may have led to limited exploration of higher doses in key clinical trials (7). Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2 65 New studies of dose– and concentration–response relations with modern methods in the past decade have led to re-evaluation of RIF dosing schedules. In preclinical models, “humanized” levels of pharmacokinetics (PK) exposure do not maximize elimination of bacilli (8). Early-phase trials have also shown incremental dose–response for RIF beyond the current dose range (9), which have been confirmed in two more recent phase-IIB trials of pulmonary disease (10, 11). Modestly higher doses of RIF are also being evaluated in treatment of extrapulmonary disease, particularly meningeal TB (12, 13), in which to date no increase has been reported in the risk of serious side-effects such as hepatotoxicity at these doses. Phase-III trials of a higher dose of RIF in pulmonary TB are under way (TRUNCATE-TB, RIFASHORT). A phase-III trial of a higher doses of RIF in disseminated HIV-associated TB is planned, with the start of recruitment towards the end of 2021 (NEW-STRAT TB, personal communication). Modern first-line regimens rely on weight-based dosing strategies to control variations in PK and, after a prolonged period of pharmaceutical development, have been successfully co-formulated as fixed-dose combination tablets by a number of manufacturers (14). While truly individualized dosing approaches have been advocated, with therapeutic drug monitoring to identify individuals at the lower end of the distribution of PK, such strategies have not been widely implemented, even in settings with the required technical resources (15). For national TB programmes globally, altered dosing recommendations for all TB patients or for selected subgroups are the only ones that could be implemented immediately worldwide. For that reason, this review focuses on evidence suitable for universal or stratified rather than individualized dosing recommendations. 1.3 How the intervention might work RIF is the key determinant of outcomes in the current first-line regimen. Its outstanding activity reduces the rate of relapse, even at shorter durations of therapy. The pharmacological basis of this critical role remains incompletely understood. It may involve several factors. RIF has particular activity against non-replicating and antibiotic-tolerant M. tuberculosis organisms in preclinical models (16) and has also been shown to accumulate in pulmonary TB lesions with repeated dosing (17). For this reason, it has been postulated that it is the key drug in eliminating a putative subpopulation of “persister” organisms, particularly in the continuation phase of treatment (18). The response to RIF-based regimens may vary by markers of disease severity, such as baseline bacillary burden and radiological evidence of cavitation (19), suggesting that people with favourable prognostic factors may have a smaller “persister” subpopulation and could be cured with a shorter regimen. Plasma concentrations of RIF vary widely among patients and between studies (20), due mainly to absorption and bioavailability, while mycobacterial killing is believed to be driven by the parameter area under the curve of a person’s RIF plasma concentrations divided by the minimum inhibitory concentration of the organism with which they are infected (AUC/MIC) (6). AUC/MIC values achieved after current doses of RIF vary by more than three times (11) and are on average much lower than those predicted to be optimal in preclinical systems, especially in the cerebrospinal space (21). The evidence presented above from dose- ranging early-phase clinical trials supports the concept that, by overcoming these issues, RIF could plausibly increase the rates of treatment success at the current duration of treatment and/ or reduce the duration necessary to achieve clinically acceptable results. Whether these possible benefits can be realized in terms of long-term outcomes without important additional toxicity has not, however, been established. Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 266 1.4 Why this review is important Several lines of evidence, summarized above, suggest that strategies for dosing RIF as part of the current first-line regimen for treatment of TB may not be optimal. Ensuring that the efficacy of the regimen is maximized could be important in improving the long-term outcomes of TB treatment for all patients and making it more robust to variations in adherence, PK and pharmacogenetics and to the emergence of resistance. Furthermore, intensification of treatment with higher doses could be important for people with severe or disseminated disease or in populations particularly vulnerable to increased variation in PK or drug–drug interactions. Higher doses of RIF could also ultimately reduce the duration of first-line treatment to less than 6 months, reducing the burden on the health system of treatment supervision and support and possibly better preventing the emergence of resistance to RIF. Objective To assess the efficacy and safety of doses of RIF higher than those currently recommended by WHO when used as part of a combination regimen for treating people with presumed drug- susceptible TB. 1.5 Methods The review was conducted according to the principles outlined in the PRISMA statement and PRISMA-P checklist (22, 23). Criteria for considering studies for this review Types of studies Randomized controlled trials Types of participants Inclusion criteria • Adults being treated for TB for the first time or being re-treated after a previous episode of TB Adults are defined as people aged ≥ 18 years, or treated as adults at participating centres in trials (inclusion of participants < 18 years of age to be clearly described). All forms of TB (pulmonary, extrapulmonary and disseminated) were considered for the purposes of the review, whether the diagnosis was based on bacteriological confirmation, imaging, biopsy or a clinical decision to treat. Exclusion criteria • People being treated for TB with confirmed resistance to RIF, INH, PZA or ETH Types of interventions Intervention • Anti-TB treatment regimens containing RIF at doses higher than those recommended in current WHO guidelines (8–12 mg/kg) Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2 67 Comparator • Anti-TB treatment regimens containing RIF at doses recommended in current WHO guidelines (8–12 mg/kg) The intervention was defined on the basis of comparisons based on dose, not on target concentrations in plasma, and randomized concentration–controlled trials were not included in the review. Although not recommended by WHO, administration of RIF by the intravenous route as an alternative to oral was permissible, provided that the target dose was different from those currently recommended. For the purposes of this review, the dose metric that was used was actual weight-adjusted doses expressed in mg/kg. It was expected that many studies would not specify the target weight- adjusted dose in the report. When absolute doses were specified, the weight-adjusted dose for each trial was determined when possible from the weight data provided in the study report or, if these data were not available, the average weight in all the trials included in the analysis. When weight-banded dosing regimens were used, the average target weight-adjusted dose stated in the study report or the manufacturer’s summary of product characteristics was accepted, or, if not specified, the average of weight-adjusted doses computed from the mid-point of each band. Comparisons according to dose were considered between regimens of similar composition and duration. Regimens of any duration – whether greater than or less than 6 months – were considered potentially eligible. Direct comparisons between regimens of different doses and durations were not considered in the primary analysis. Types of outcome measures Primary outcomes We used definitions consistent with WHO programmatic outcomes (24) but adapted to permit evaluation of regimens longer or shorter than 6 months and of follow-up data in clinical trials and of trials of extrapulmonary TB. Treatment success In adults with pulmonary TB (bacteriologically confirmed, presumed drug-sensitive): smear or culture negative in the last month of treatment and on at least one previous occasion OR completed treatment without evidence of failure BUT with no record to show that sputum smear or culture results in the last month of treatment and on at least one previous occasion were negative, either because the tests were not done or because results are unavailable. In adults with extrapulmonary TB: resolution of clinical symptoms and signs of TB at the end of treatment as judged by the investigators. Treatment failure In adults with pulmonary TB (bacteriologically confirmed, presumed drug-sensitive): sputum smear or culture positive within the last month of treatment. In adults with extrapulmonary TB: failure to resolve or return of clinical signs and symptoms of TB by the end of treatment as judged by the investigators. Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 268 Relapse In adults with pulmonary TB (bacteriologically confirmed, presumed drug-sensitive): smear or culture positive on one or more occasions during a defined period of follow-up after having previously met the criteria for treatment success. In adults with extrapulmonary TB: return of clinical signs and symptoms of TB as judged by the investigators after having previously met the criteria for treatment success. Death Death from any cause before starting or during the course of treatment or during follow-up. Adverse events All reported adverse events expressed as number of participants experiencing an event. Secondary outcomes Serious adverse events All adverse events detailed in the study report that resulted in death, were life-threatening, required hospitalization or prolongation of hospitalization, resulted in persistent or significant disability or incapacity, consisted of a congenital anomaly or birth defect or were another important (protocol- defined) medical condition. Discontinuation Adverse events resulting in discontinuation of study medication. Drug-specific adverse events of interest Drug-induced liver injury Rifamycin hypersensitivity syndromes (thrombocytopenia, influenza-like illness) Disease-specific efficacy outcomes of interest Outcomes specific to trials in which people with extrapulmonary forms of TB were recruited, such as neurological disability in meningeal TB and pericardial constriction in pericardial TB. Search methods for identification of studies We attempted to identify all relevant studies regardless of language, date of publication or publication status (published, unpublished, in press or in progress). The search strategy is outlined in Annex 1. Electronic searches We searched the following electronic resources for eligible trials: MEDLINE, EMBASE, CENTRAL (Cochrane central register of controlled trials), Cochrane Infectious Diseases Group clinical trials register, WHO International Clinical Trials Registry and Clinicaltrials.gov We used the search strategy shown in Annex 1 (for CENTRAL, MEDLINE and EMBASE), which consists of the Cochrane Highly Sensitive Search Strategies for identification of clinical trials appropriate to those resources (25). Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2 69 Search for other resources We searched the reference lists of the retrieved study reports for any other studies. Active investigators in the field were contacted to provide information on unidentified, ongoing or planned trials. Results of searches were also cross-referenced to the existing database of clinical trials of first-line TB treatment generated by our group for the systematic reviews conducted by the PreDICT-TB consortium (25)1. Data collection and analysis Selection of studies After removal of duplicates, the titles and abstracts of unique studies identified in the search results were screened independently by two authors and compared with the inclusion and exclusion criteria. Full texts were retrieved for all citations considered “eligible” or “unclear” but not for those considered “ineligible”. Two authors independently screened the retrieved full-text study reports, marked them as “eligible” or “ineligible” and recorded the reasons for exclusion in each case. When appropriate, several reports of the same study were collated. Disagreements were resolved by discussion, contact with the study investigators or with the assistance of a third author. The Covidence web-based interface was used to manage these processes (26). Title and abstract and full text screening were undertaken for the four main anti-TB drugs (RIF, INH, PZA and ETH) and separated into drug-specific studies at the final stage. Data extraction and management We designed and pilot-tested a data extraction form and modified the form according to the results of the pilot test. Two authors independently extracted data and compared the results. A few discrepancies were resolved through discussion and comparison with the study reports. The data extraction form included: Source: lead author, year of publication, journal, PubMed ID, sponsor, funding source, trial registration number, corresponding author Methods: study design, study duration, sequence generation, allocation concealment, blinding, completeness of outcome data, selective reporting, other concerns about bias Participants: number recruited, country or countries, patient status (new or retreatment), setting, inclusion and exclusion criteria, age, sex, HIV status, primary prophylaxis and availability of antiretroviral treatment, diabetes mellitus status, nutritional status Microbiological methods: details of diagnostic and susceptibility testing methods Pharmacology: availability of data on PK and/or pharmacogenetics Interventions: number of arms, names of drugs, doses of drugs, frequency of dosing, use of fixed-dose combinations, duration and structure of regimen, use of direct observation Outcomes: for each review outcome, where available, we extracted the number randomized to each arm, the number included in the analysis and the number that experienced the event in order to conduct intention-to-treat (ITT), per-protocol (PP) and complete case analyses. Extracted data were imported into R version 4.0.4 for further management and analysis (27). 1 Ken-Dror G, Bonnet L, Koh G, Davies G (2021) Comparing the efficacy of first-line drug regimens for pulmonary tuberculosis: meta- analysis of Phase III Trials or pulmonary tuberculosis: meta-analysis of phase III trials (personal communication) Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 270 Assessment of risk of bias Two authors independently assessed the methodological quality of each included study using the Cochrane “risk of bias” tool. The results are reported in a table (28). We resolved any disagreements through discussion. Regarding generation of allocation sequence and allocation concealment, we classified each as adequate, inadequate or unclear in each included study (29). We reported who was blinded in each included study and assessed the risk of bias associated with blinding separately for each primary outcome. If at least 90% of participants were followed up until study completion, we classified inclusion of all randomized participants as adequate; otherwise, we classified inclusion as inadequate. We attempted to contact the study authors if information was unspecified or unclear. Measures of treatment effect We used risk ratio (RR) as the measure of treatment effect in analyses of both efficacy and safety and also expressed results as risk difference (RD) where appropriate. Missing data Primary outcomes were analysed on an ITT basis, with missing outcomes assumed to be adverse. As only one trial reported significant (> 10%) loss to follow-up, we did not conduct a sensitivity analysis. Assessment of heterogeneity We assessed heterogeneity by visual inspection of forest plots to determine the closeness of point estimates to each other and overlap of confidence intervals (CIs). For direct comparisons and meta-analyses, we used the I2 statistic to assess heterogeneity, with a value of 50% taken to indicate significant statistical heterogeneity. Assessment of reporting biases We conducted visual inspection of funnel and Galbraith plots of the studies for any evidence of publication bias. Data synthesis The review prioritized direct within-trial comparisons between doses of the individual drugs as part of otherwise similar regimens at the same total duration of treatment, to the extent permitted by the data. Meta-analyses were performed as supported by the structure of the dataset. Direct comparisons Meta-analysis, when appropriate, was performed with random effects models according to the DerSimonian-Laird method (30). When we wished to include more than one intervention study arm from a multi-armed study, we compared each intervention arm separately with the comparator arm to avoid splitting the control group. Stratified forest plots were used to present the data graphically. Corresponding 95% CIs and P values were computed with a significance level of 0.05. We qualified heterogeneity with the I2 statistic. We were unable to complete meta- regression as planned, with dose as a covariate to characterize dose–response relations for RIF owing to insufficient data with comparable background regimens between studies. Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2 71 Indirect comparisons We were unable to complete network analyses, as we could not form a coherent network. Statistical analyses were conducted with R package metafor (27). Subgroup analysis and investigation of heterogeneity We were unable to complete any subgroup analysis owing to the paucity of data. There were insufficient data for the planned subgroup analyses, including: first versus retreatment, severity of disease, diabetes, HIV infection, companion drugs in the regimen, pharmacogenetics polymorphism, pulmonary versus meningeal TB. Heterogeneity was assessed by visual inspection of forest plots and from the I2 statistic. Certainty of the evidence We assessed the certainty of the evidence with the GRADE approach (31). We constructed “summary of findings” tables to present ratings of the certainty of the evidence for effect estimates for each outcome, with relative and absolute measures of effect. To express imprecision, we computed the “optimal information size” for each outcome and considered the width of the CI as well as the total number of observed events (32). Sensitivity analysis We performed the following sensitivity analyses as appropriate: ITT, in which all the participants who were randomized are included, and per-protocol (PP), in which only those who were randomized and completed the treatment course as planned are included. Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 272 2. Results 2.1 Description of studies Results of the search The searches identified 4245 records. Of these, we excluded 1788 duplicate records. Of the remaining 2357, we excluded 1763 after assessing the titles and abstracts. We retrieved 107 full-text studies to assess their eligibility for inclusion. Fig. 1 shows the screening process in a PRISMA flow diagram. Fig. 1. PRISMA diagram Search N=4245 Title/Abstract Screening N=2457 Included in Qualitative Synthesis N=13 Included in Quantitative Synthesis N=13 Full-text Retrieval N=107 Eligible RIF N=22 Eligible RIF, INH, PZA or ETH N=38 Duplicates N=1788 Excluded N=2350 Excluded N=69 33 wrong intervention 30 wrong comparator 3 wrong study design 2 wrong population 1 duplicate Excluded N=9 2 different route same dose 3 PK substudies of already included studies 1 protocole only 3 duplicate Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2 73 Included studies We included 13 studies with a total of 4411 participants. Details are provided in Annex 2. All but one of the studies was in full format. That by Merle et al. (33) was presented orally and is yet to be published. Geographical location and period The studies were conducted at sites in Bangladesh, Benin, Bolivia (Plurinational State of), Guinea, Indonesia, Nepal, Peru, Senegal, South Africa, Uganda, United Republic of Tanzania, United States of America (USA) and Viet Nam. With the exception of the study in the USA, all the studies were conducted in low- or middle-income countries, where the burden of TB disease is greatest (1). Participants were recruited in 1970–1972 in one study and, when dates were provided, between 2000 and 2016 in all the other studies. Participants The review criteria for inclusion specified adult participants; however, the definition of “adult” differed by country. In our studies, although most (n=8) studies included participants aged ≥ 18 years, some included participants aged ≥ 14 years (34), ≥ 15 years (35–37) and ≥ 17 years (38). As it was not possible to determine clearly which participants were > 18 years in these trials, all participants were included, on the understanding that the participants in the studies represented adults (> 18 years of age) or those treated as adults as defined by the study authors. Eight studies recruited participants with pulmonary TB; four studies had participants with tuberculous meningitis (13, 35, 36, 38), and one study included patients with any form of TB (32). HIV status was documented in all but one study (34). Two studies recruited only participants with HIV (33, 39). In one study, all participants were not tested, but known HIV-positive individuals were excluded (37). The other nine studies appear to have tested every participant. The proportion of participants who were HIV co-infected in the other trials varied from 0 to 43%. Antiretroviral treatment was clearly documented as available to study participants in only four studies (33, 35, 39, 40). Interventions Twelve trial reports expressed doses of RIF as mg/kg body weight. Long et al. (34) specified doses in mg (450, 600 or 750), and mg/kg values were estimated from the weight-band data provided in the report. The doses evaluated varied from 8 to 35 mg/kg. Eleven trials included a standard dose of 10 mg/kg for comparison with higher doses. In one, the lowest dose was 13 mg/kg given intravenously (38). In another study, an oral dose of 10 mg/kg was compared with an IV dose of 13 mg/kg (35). We did not include data from comparisons of oral and IV regimens in our meta-analyses but reported these separately where appropriate. As the background treatment regimens had to be identical for the purposes of meta-analysis, data from a number of studies could not be compared directly. The reasons included addition of levofloxacin (40) and second randomization to moxifloxacin (10) or ethambutol (35). These studies therefore contributed data only to analyses of selected outcomes, and the interpretation was qualified by consideration of these additional components of the intervention. Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 274 Participants were recruited as inpatients for all four studies of tuberculous meningitis (35, 36, 38, 40). Both inpatients and outpatients were recruited in three studies (10, 34, 41) and only outpatients in three other studies (11, 12, 37). Of the remaining three, one would appear to have recruited only outpatients (42), although it is not clearly stipulated, and it is not clear where recruitment and treatment were done in the other two studies (33, 39). Treatment was directly observed during the intervention period in eight studies. Treatment was directly observed for half of the intervention period in one study (42), not observed in two studies (10, 34) and unclear as to whether it was observed in the remaining two studies (12, 33). Follow-up The length of follow-up varied considerably among studies, from 2 weeks (38) to “a minimum of one year” (34), 18 months (33) and 1 year after cure or completion of treatment (37). Outcome measures Primary outcomes Eight studies provided data on treatment success (10, 11, 33, 34, 37, 39, 41, 42). Various definitions of treatment success were used by the investigators, which differed from the WHO definition adopted for our protocol. The definitions included: • culture conversion on Lowenstein-Jensen (LJ) medium at 12 weeks • culture conversion on LJ and mycobacterial growth indicator tube (MGIT) at 8 weeks • culture conversion on solid media at 26 weeks • culture conversion at 8 weeks • culture conversion at 20 weeks • culture conversion 12 months after cure or treatment completion • not specified (oral presentation) • culture conversion at 6 months • initially smear positive culture converted in the last month and on at least one previous occasion Seven studies provided data on treatment failure (12, 33, 34, 37–39, 41). Four studies presented data on relapse (11, 32, 34, 37). The minimum follow-up time in these studies was 1 year. Many other studies appropriately did not report relapse rates, as most followed up participants to ≤ 6 months, during which time they would have been receiving standard anti-TB treatment and thus collection of data on relapses would not have been feasible. All but one study (38) provided data on deaths. All the studies reported some information on adverse events, with wide variation in how data were presented: as the total number of adverse events or the total number of participants who had at least one adverse event. A number of studies reported only hepatotoxicity. Reporting on adverse events was the most consistently reported outcome. Secondary outcomes Six studies provided defined data on serious adverse events (SAEs) (10, 11, 37, 39, 41, 42). Discontinuation was reported in only four studies (10, 34, 35, 38). Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2 75 All but one study (34) reported the drug-specific adverse event of drug-induced liver injury with RIF. Most authors did not, however, specify whether they considered that the reported drug- induced liver injury was due to RIF or to other potentially hepatotoxic study drugs, such as isoniazid and pyrazinamide. Disease-specific efficacy was not reported consistently. The four studies of participants with tuberculous meningitis reported neurological outcomes variably, but follow-up was frequently for a short time, and the authors placed more emphasis on mortality. Excluded studies The searches, title and abstract and full-text screening were all initially completed together for randomized controlled trials (RCTs) with direct comparisons of doses of each of the first line anti-TB drugs (RIF, INH, PZA, ETH). During the initial full text screening, 68 studies were excluded: 33 for the wrong intervention, 29 for the wrong comparator, 3 for the wrong study design and 2 for the wrong patient population, with 1 duplicate. At this stage, studies were identified by the drug for which a direct dose comparison was described. Of the 39 studies then included, 23 were identified as possibly including a direct dose comparison for RIF. On further full-text review, 10 more studies were excluded: 3 were duplicates (including posters and oral presentations of included full papers), 2 reported on IV versus oral routes of administration of the same doses, 3 were on PK for included full RCT reports, 1 was a protocol and in 1 a standard dose of RIF was compared with a lower dose. The reasons for excluding studies are summarized in Annex 3. Assessment of risk of bias We assessed the risk of bias in the included studies with the Cochrane risk of bias assessment tool (43). We separated the risk of bias in efficacy and safety outcomes and generated tables accordingly (Figs 2 and 3) using the risk-of-bias visualization tool (44). Fig. 2. Risks of bias in efficacy outcomes Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 276 Fig. 3. Risks of bias in safety outcomes Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2 77 Allocation The majority of the studies had a low risk of bias for random sequence generation, as a number of methods were clearly described. Three studies had an unclear risk of bias, as the procedures were not clearly described (12, 33, 34). Most of the studies also described adequate allocation concealment, resulting in a low risk of bias. Five studies did not, however, report such procedures, and the risk of bias was unclear (10, 12, 33, 34, 36). Blinding Five studies reported double-blinding procedures (11, 12, 36, 40, 41), resulting in a low risk of performance bias. All but one (36) of these five studies reported concealment of the treatment group from outcome assessors, resulting in a low risk of detection bias. The other double- blinded studies did not clearly state whether the outcome assessors were blinded; thus, the risk of detection bias was unclear. Seven studies described an open-label procedure in their RCTs (10, 33, 34, 37–39, 42), resulting in a high risk of performance bias. Two of these studies (10, 42) blinded their laboratory personnel, resulting in an unclear risk of detection bias for safety outcomes and a low risk of bias for efficacy outcomes, as both were studies of pulmonary TB and thus laboratory personnel would have provided data on efficacy. One study was that presented orally (33) and therefore did not contain as much detail as the full papers; for this study, the risk of detection bias was unclear. The other five studies were fully open label and therefore with a high risk of detection bias. Blinding of outcome assessment for safety would be impacted by an open-label design; however, for studies of tuberculous meningitis in which outcome data were collected mainly for mortality, blinding of outcome assessment would not affect efficacy data, resulting in a low risk of bias. Incomplete outcome data All but one study reported > 90% follow-up of participants; therefore, the risk of attrition bias was low. The reasons for withdrawal were not documented in the study that reported < 90% follow-up of participants (10), and therefore the risk of attrition bias was high. Selective reporting Most of the studies showed no evidence of selective reporting. In one study, the “safety and tolerability” of the regimen was a secondary outcome, but no SAE data were provided (36). A study in which tolerability was assessed as a primary outcome reported no SAEs above a certain concentration of RIF and reported no difference among groups in the discussion, but Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 278 did not provide the SAE data (12). A study with safety and tolerability as a primary outcome did not provide any SAE data and reported that “adverse events (e.g., new neurological signs) were not incorporated in the assessment of safety and tolerability” (38). The risk of selective reporting bias in these three studies is unclear. Publication bias Publication bias was assessed from funnel and Galbraith plots for meta-analyses of primary outcomes incorporating three or more trials (treatment success, mortality and adverse events) (see Annex 4). Visual inspection of these plots revealed no clear evidence of publication bias. Other potential sources of bias A few participants were randomized in error in one study (10). This could have led to bias. In another study, the inclusion criteria were changed during the trial, to reduce the lower weight limit and remove a CD4 count limit for participants (39), which could also have led to bias. In one study, an error in one of the intervention groups was noted, whereby one weight band received fewer fixed-dose combination tablets than they should have and a higher extra dose of RIF; they would therefore have had a lower background dosage each of isoniazid, pyrazinamide and ethambutol (37). This could have led to performance bias. 2.2 Effects of interventions Forest plots summarizing the ITT and PP analyses of all outcomes and comparisons are presented in Annex 5. Treatment success Of the eight trials that provided data on treatment success with various definitions, data synthesis with a measure that fully or partially met our protocol-defined end-point could be performed for four studies that reported use of 10 mg/kg versus 15 mg/kg (11, 33, 41, 42) and for six studies that reported use of 10 mg/kg versus 20 mg/kg (10, 11, 37, 39, 41, 42). There was no evidence of higher rates of treatment success at doses of RIF of 15 mg/kg (RR 1.00 [0.97 ; 1.03], RD –1.0% [–5.5 ; 3.5], 4 trials, 916 participants) or 20 mg/kg (RR 1.01 [0.97 ; 1.06], RD 2.9% [–1.1 ; 7.0], 6 trials, 1302 participants). Similar results were obtained in PP analyses (Annex 5, analyses 1–3). Boeree et al. (10) reported a similar rate of treatment success at doses of 10 and 35 mg/kg of RIF (RR 0.98 [0.91 ; 1.06], 1 trial, 186 participants). These data were included in the ITT but not in the PP analysis because more participants were reported as having culture-converted in one group (RIFQHZ) than the number of participants included in that analysis in total in their published mITT analysis (Table 2 in the study report). Treatment success data from Merle et al. (33) were also included in the ITT analysis but not in the PP analysis. The results of this study were available only as an oral presentation and PP denominator values were not available. Long et al. (34) used dose levels distinct from those used in the other included studies, and we elected not to combine the data. The trial suggested that doses < 9 mg/kg RIF would not be effective, but no statistically significant difference was found between the estimated 10 mg/kg and 13 mg/kg doses (based on weight bands provided as before). Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2 79 Treatment failure The definition of treatment failure that we used was sputum smear or culture positive within the last month of treatment for pulmonary TB or failure to resolve or return of clinical signs and symptoms by the end of treatment for extrapulmonary TB. In the seven studies that reported treatment failure, the definitions used by the investigators were inconsistent. For example, one study reported treatment failure, but participants were followed up only to 12 weeks (41), by which time our definition of treatment failure could not have been reached. Another paper reported one episode of treatment failure, although the participant was found to have multi- drug resistant (MDR) TB on bone marrow examination at 6 weeks into treatment (41) and would thus be excluded from our review of drug-susceptible TB. A study in which only people with pulmonary TB were recruited reported treatment failure when participants were judged to be clinically worse by the study doctors, although sputum cultures from most of these participants “remained negative” (34). In two studies (35, 37), the definition of treatment failure was in line with ours, based on sputum smear positivity from month 5 onwards. Two other studies (11, 33) did not define treatment failure but are recent studies of patients with pulmonary TB and are assumed to have met the definition we used, as per WHO recommendations. These studies were sufficiently similar to permit data synthesis, except for that of Ruslami et al. (12), in which doses of 10 mg/kg and 13 mg/mg were compared. Insufficient data were available from the study of Merle et al. (33) to conduct a PP analysis. The evidence suggests that RIF at a dose of 15 mg/kg reduces treatment failure (RR 0.71 [0.27 ; 1.88], RD –1.0% [–3.6 ; 1.6%], 2 trials, 616 participants), while a dose of 20 mg/kg makes no difference (RR 1.01 [0.29 ; 3.61], RD 0.5% [–1.9 ; 2.8], 2 trials, 821 participants) (Annex 5, analysis 4). Relapse Four trials reported relapse rates. Long et al. (34) reported on relapses, but the data could not be reliably extracted as the denominators in the report were inconsistent. In two studies (11, 37), the definition of relapse was not consistent with our protocol. Merle et al. (33) did not define relapse. In one paper, doses of 10 mg/kg and 13 mg/mg were compared; however, the others were used in the meta-analysis (Annex 4, analysis 5). Only an ITT analysis was performed, as Merle et al. (33) did not provide sufficient data for a PP analysis. RIF at a dose of 15 mg/kg (RR 0.79 [0.25 ; 2.45], RD –1.6% [–4.1 ; 0.7%], 2 trials, 617 participants) or 20 mg/kg (RR 0.93 [0.07 ; 12.72], RD 0.0% [–1.2 ; 1.2], 2 trials, 821 participants) may slightly reduce the risk of relapse (Annex 5, analysis 5). Death Most trials reported all-cause mortality as an outcome. The data were comparable in most of the studies and were combined for ITT and PP analyses (Annex 4, analyses 6–9). The evidence suggests that RIF at a dose of 15 mg/kg (RR 0.80 [0.47 ; 1.37], RD 1.1% [–4.0 ; 1.8], 4 trials, 916 participants) or 20 mg/kg (RR 0.93 [0.47 ; 1.81], RD 0.5% [–2.2 ; 1.3] probably reduces all-cause mortality slightly (7 trials, 1341 participants). In two studies, doses of RIF higher than 20 mg/kg were compared with control (10 mg/kg), but the evidence in both is very uncertain because of the very small numbers of participants or events. Dian et al. (36) Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 280 found that a dose of 30 mg/kg may reduce all-cause mortality as compared with control (RR 0.43 [0.13 ; 1.43], RD –20.0% [–46.1 ; 6.1%], 1 trial, 40 participants). Boeree et al. (10) found that a dose of 35 mg/kg of RIF may increase all-cause mortality (RR 5.81 [0.24 ; 140.66], RD 1.6% [–1.5 ; 4.6%], 1 trial, 186 participants) as compared with control (10 mg/kg). Adverse events Each study reported adverse events to some extent. Six studies reported the total numbers of several adverse events rather than the number of participants who had an adverse event. These studies could not be compared in a meta-analysis as the background regimens varied widely. None of the studies found a statistically significant difference in adverse events among groups. A number of studies also reported adverse events that were side-effects of other components of the background regimen, such as SQ109 and fluoroquinolones, which could not be compared. Serious adverse events Six studies provided data on the number of participants with SAEs, not including deaths. Little or no increase in the number of SAEs was seen with RIF at a dose of 20 mg/kg (RR 0.96 [0.53 ; 1.76], RD 0.6% [–3.0 ; 1.7], 6 trials, 1310 participants), while the estimate for RIF at a dose of 15 mg/kg (RR 1.83 [0.5 ; 6.69], RD 1.4% [–1.4 ; 4.2]; 3 trials, 417 participants) suggests that higher doses might increase the number, although the evidence is very uncertain (Annex 5, analysis 10). Boeree et al. (10) provided data on SAEs seen at 10 mg/kg and 35 mg/kg, which suggest that the higher dose might have increased the number of SAEs (RR 1.30 [0.38 ; 4.44], 1 trial, 186 participants), but the evidence is very uncertain. Discontinuation Only four studies reported discontinuation. The data were not comparable because different background regimens were used. Few participants were reported to have discontinued treatment in any of the studies. Drug-specific adverse events Drug-induced liver injury All but one study reported drug-induced liver injury as an outcome, predominantly with grade-3 or -4 hepatotoxicity as the marker. Long et al. (33) reported various adverse events, including changes in transaminase activity, but these were not consistent enough for statistical evaluation. Little or no difference was found in the risk of drug-induced liver injury with RIF at a dose of 15 mg/kg (RR 0.99 [0.54 ; 1.83 ], RD 1% [–2.8 ; 5.0], 3 trials, 420 participants), but the evidence is very uncertain. For RIF at 20 mg/kg (RR 0.99 [0.63 ; 1.56], RD 0.0% [–2.3 ; 2.3], 6 trials, 1310 participants), the evidence suggests little or no difference. Higher doses than 20 mg/kg were compared with control (10 mg/kg) in two studies, but the evidence was very uncertain in both. Dian et al. (35) compared a dose of 30 mg/kg with control (RR 1.33 [0.34 ; 5.21], RD 5.0% [18.5 ; 28.5], 1 trial, 40 participants), while Boeree et al. (10) compared a dose of 35 mg/kg of RIF with control (RR 5.86 [0.62 ; 55.17], RD 3.9% [–1.5 ; 9.4]1 trial, 190 participants) (Annex 5, analyses 11–13). Thrombocytopenia Six studies reported data on thrombocytopenia. The data were not directly comparable because of different doses and background regimens. Aarnoutse et al. (40) reported no episodes of thrombocytopenia at 10 or 20 mg/kg but two episodes at 15 mg/kg. Atwine et al. (38) reported Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2 81 four episodes of thrombocytopenia at 10 mg/kg, two at 20 mg/kg with 600 mg efavirenz and three at 20 mg/kg with 800 mg efavirenz. Dian et al. (35) reported two episodes of thrombocytopenia at 10 mg/kg, four at 20 mg/kg and two at 30 mg/kg. Heemskerk et al. (39) also showed no difference in thrombocytopenia rates, reporting 17 episodes at 10 mg/kg and 14 at 15 mg/kg (plus levofloxacin). Long et al. (34) reported data inconsistently but reported sustained thrombocytopenia in 28 patients at 8 mg/kg, 17 at 10 mg/kg and 39 at 13 mg/kg. Ruslami et al. (35) reported no episodes in any arm. Rifamycin hypersensitivity syndrome Rifamycin hypersensitivity syndrome was discussed in five studies. Four reported no episodes of hypersensitivity (11, 12, 34, 41), while Ruslami et al. (35) reported three episodes of hypersensitivity, all at the higher dose (13 mg/kg intravenous), with none in the control (10 mg/ kg orally). Two of the episodes were described as mild. In the third, however, the patient had an anaphylactic reaction and died directly after receiving RIF and moxifloxacin intravenously; it was not apparent which agent caused the anaphylaxis. The studies in which hypersensitivity was reported included 1262 participants, suggesting a crude rate of < 0.003%. Disease-specific efficacy outcomes Participants with tuberculous meningitis were recruited in four studies. One study included neurological disability at 9 months and the time to the first new neurological event or death as secondary outcomes (40); another study (35) included neurological outcomes as a secondary outcome. Data on neurological outcomes were not collected in the other two studies (36, 38). Heemskerk et al. (40) found little or no difference between the comparator (10 mg/kg RIF) and the intervention (15 mg/kg RIF and 20 mg/kg levofloxacin) (RR 1.03 [0.86 ; 1.24], RD 1.1% [–5.5  ; 7.6%],1 trial, 817 participants). Ruslami et al. (35) found that a higher dose of RIF (13 mg/kg IV) may have improved complete neurological recovery after 6 months of treatment as compared with the lower dose (10 mg/kg orally) (RR 2.41 [0.83 ; 6.97], RD 18.1 [–2.4 ; 38.7%], 1 trial, 60 participants), but the evidence is very uncertain. Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 282 3. Discussion 3.1 Summary of main results This review addressed the question of whether doses of RIF higher than 10 mg/kg could safely improve the outcomes of first-line treatment of TB in adults. Thirteen trials met the inclusion criteria and, although the studies had diverse designs and reported outcomes, meaningful data synthesis was possible for the objectives of the review. Participants with pulmonary or meningeal TB were recruited for all the trials. The doses of RIF represented in the review ranged from 8 to 35 mg/kg, with some variation in the companion drugs in the background regimen. All the trials of higher-dose RIF were published within the past 15 years, with the notable exception of Long et al. (1979) (34). A tendency towards a slow increase in the dose of RIF from the currently recommended 8–12 mg/kg to doses of 13 mg/kg, 15 mg/kg and 20 mg/kg was seen. The two studies of higher doses – 30 mg/kg (36) and 35 mg/kg (10) – were both published within the past 4 years. Higher doses of RIF were not associated with more successful treatment. There was some evidence that they might reduce treatment failure, relapse and all-cause mortality, but these outcomes were uncertain, and the observed dose–response relations were inconsistent. Most of these data, however, were from trials of doses up to 20 mg/kg, and the review exclusively addressed programmatically important outcomes. Many of the trials were designed as early- phase studies and therefore had insufficient power individually to measure these outcomes or did not report them, particularly the critical relapse outcome. Many authors claimed evidence of improved efficacy on the basis of the investigator-defined primary end-points, which were typically measures of culture conversion at 8 or 12 weeks of therapy. For instance, Boeree et al. (10) concluded that a dose of 35 mg/kg RIF reduced the time to culture conversion as compared with 10 mg/kg. Jindani et al. (42) reported higher culture conversion with 20 mg/kg than with 10 mg/kg, although the increase was not statistically significant, and Velasquez et al. (11) found a dose–response relation at 15 and 20 mg/kg in statistical modelling of sputum colony counts. These intermediate outcomes were not, however, in the scope for this review, and it is uncertain whether improvements in culture status at early times translate consistently into better long-term outcomes. On the basis of the available evidence, summarized in this review, this has not been demonstrated for higher doses of RIF. Although reporting on safety imposed some limitations on data synthesis, and interpretation of adverse events was complicated by the diversity of the background regimens in which RIF was used, it appears that higher doses of RIF, at least those up to 20 mg/kg, may not increase the rates of SAEs or rifamycin-specific adverse effects such as drug-induced liver injury, thrombocytopenia or hypersensitivity syndromes. The evidence on safety in the two studies at 30 mg/kg and 35 mg/kg of RIF was very uncertain, although the point estimates suggested a higher risk of drug- induced liver injury. Some of the adverse events of interest are known to be relatively rare, and the size of the available dataset did not allow reliable exclusion of an increase in events of such low frequency, although our review suggests that the incidence of rifamycin hypersensitivity is very low. This meta-analysis does provide some reassurance that the risk of hepatotoxicity, which may occur at an incidence of ≥ 1% during first-line therapy and has been a frequent rationale for caution in escalating the dose of RIF (7), may not be increased at doses up to 20 mg/kg; however, additional data on higher doses are required. Discontinuation of therapy was Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2 83 infrequently reported, but the rates were low. Similarly, all-cause mortality was probably slightly lower with higher doses of RIF up to 20 mg/kg. A number of the trials were either primarily studies of PK or included nested PK studies. All but one of these (38) reported that higher doses of RIF reliably resulted in higher exposure in plasma and/or cerebrospinal fluid, supporting the PK–pharmacodynamics rationale for the intervention in previous preclinical and clinical studies (12, 21, 36, 41). Studies not included in this review suggest that still higher PK exposures may be achievable and tolerable in humans at doses of up to 40 mg/kg, suggesting that there remains some scope for increasing the dose of RIF in future trials (45). 3.2 Overall completeness and applicability of evidence Most of the trials included in this review were conducted relatively recently and examined changes in RIF dosing with the companion drugs that comprise the current first-line regimen. Although some trials included more than one higher dose of RIF, allowing evaluation of dose– response relations within and between trials, many had only one intervention arm. The trial interventions were exclusively for pulmonary and meningeal TB, while the forms of TB for which there may be the strongest rationale for intensified RIF dosing are disseminated or other forms of extrapulmonary TB. The dose levels used by the investigators in the trials in patients with tuberculous meningitis were lower than those investigated for pulmonary TB. The settings of the studies were diverse in terms of both geographical area and TB incidence. The RCTs represented many of the countries with the highest incidence rates of TB but did not include China or India. We were unable to identify the data in some of the studies for patients < 18 years and decided that it was appropriate to include this minority of younger patients in our analysis rather than to exclude the studies. HIV co-infected participants are relatively under-represented in this review, because many investigators imposed restrictions on use of antiretroviral therapy in order to avoid drug–drug interactions that might reduce virological efficacy and because most of the trials were conducted in settings where HIV co-infection was uncommon and/or they explicitly excluded HIV-positive patients or enrolled them only if their baseline CD4 count was > 200 cells/mm3. These restrictions limit immediate applicability of the evidence in settings where HIV co-infection is common. Although people with diabetes and TB are known to have poorer outcomes on treatment, diabetes was mentioned in only a few study reports and was not part of the rationale for any of the studies. The needs of these patients were therefore not addressed in the trials included in this review (5, 46). 3.3 Certainty of the evidence In general, as most of the trials included in this review were small phase-II studies, the evidence was well below the optimal information size for all the outcomes of interest except for treatment success. This is reflected in the broad confidence intervals, which typically included substantial benefit and harm and, in some cases, also very few events, particularly for safety outcomes. Nevertheless, heterogeneity, as measured with I2 values, was uniformly low for almost all the analyses. We summarized the certainty of evidence with the GRADE approach for all the outcomes in a “summary of findings” table (Annex 6). Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 284 We had some problem in evaluating risk of bias, particularly for safety outcomes. We classified five studies as having high overall risk of bias and five as having an unclear risk of bias, using the Cochrane risk of bias assessment tool (28). The authors of a number of the studies reported that their trials were open-label, as they considered that participants could not be blinded to dosing with RIF owing to the orange discoloration of body fluids. The authors who conducted double-blind trials used placebos and did not comment on this potential issue. The open-label trials included in this review were judged as at high risk of bias for safety assessment because, although the authors did not consider blinding to be feasible, the risks of performance and detection bias remain and could, for example, have resulted in greater ascertainment of drug- induced liver injury events in the intervention arms because investigators were concerned about this potentially serious side-effect. 3.4 Limitations As most of the trials were not designed as phase-III trials, few reported on the critical relapse outcome, and we were unable to assess a number of secondary outcomes or perform planned subgroup analyses owing to a paucity of data or unclear reporting. We were also unable to compare a number of the studies directly because of differences in background regimen, as many investigators tested novel, potentially more potent regimens rather than focusing on defining the dose–response of RIF. Although the range of doses was quite wide, most of the data were on the dose range 10–20 mg/kg. Fewer trials were conducted in people with tuberculous meningitis than with pulmonary TB, and lower doses of RIF were evaluated. As noted above, people with HIV co-infection were under-represented, and not enough data were available to draw any conclusions for other subgroups, such as those with diabetes or lower body weight. 3.5 Potential biases in the review process As the review was restricted to regimens as similar as possible to the currently recommended first- line regimen, several studies of use of higher RIF doses in intermittent regimens were considered out of its scope and therefore excluded. Such regimens typically consist of a cumulative weekly dose similar to daily regimens, and, as PK–pharmacodynamics interactions between dose size and interval for RIF are not completely characterized, these studies would have been difficult to interpret. We limited bias in our review by following the procedures outlined by the Cochrane group (43). An information specialist in the Cochrane Infectious Diseases Group (VL) completed the search, and we found no publication bias for key outcomes (Annex 4). It is unlikely that the search missed major studies, but small unpublished studies may not have been identified. Two of the review authors examined the search results, selected studies and extracted data independently to minimize bias in study selection and data extraction. 3.6 Agreements and disagreements with other studies or reviews We identified two previously published systematic reviews of high-dose RIF (47, 48); however, both included only studies on pulmonary TB. Steingart et al. (47) reported that doses of RIF higher than the standard improved culture conversion rates, noting that clinical trials were required Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2 85 to confirm efficacy and assure tolerability. Onorato et al. (48) reported that higher doses of RIF were associated with an increased rate of sputum culture conversion, particularly at doses > 20 mg/kg. No difference was observed in mortality between treatment groups, and similar rates of hepatotoxicity were observed. Although these reviews focused on intermediate bacteriological end-points rather than longer-term outcomes such as treatment success, their findings are in agreement with the conclusions of our review, particularly with regard to safety outcomes. We identified four systematic reviews of the PK of RIF. In 2016, Mota et al. (49) reported that a high proportion of patients undergoing treatment for TB received subtherapeutic drug concentrations. In 2018, Stott et al. (20) reviewed the PK of RIF in adult TB patients and healthy volunteers and also found that measures of RIF exposure were often below the accepted threshold; they advised an increase in RIF dosage to increase its efficacy. Two more recent reviews indicate that the PK of RIF appears to be related to treatment outcome and that a low concentration of RIF may increase the risk of poor outcomes (50, 51). Although these reviews are less directly relevant to our findings, they strengthen the rationale for using higher doses of RIF in individualized therapy and the possibility that higher doses for all might reduce the risk that a large subgroup of people on first-line therapy might receive target doses below the PK–pharmacodynamics thresholds. 3.7 Authors’ conclusions Implications for practice We found no clear evidence that use of doses higher than the recommended 10 mg/kg of RIF in pulmonary and meningeal TB is associated with higher rates of treatment success. Higher doses may be associated with lower rates of treatment failure, relapse and all-cause mortality, but the evidence for these outcomes is uncertain. Although most of the evidence was for doses up to 20 mg/kg, there was no consistent dose–response pattern in these outcomes up to doses of 35 mg/kg. We found no large differences in safety outcomes or tolerability at doses up to 20 mg/kg, particularly with regard to drug-induced liver injury and rifamycin hypersensitivity, but the evidence for an effect of higher doses on these outcomes is very uncertain. This review does not therefore provide support for increasing the dose of RIF for all patients in routine practice, and the limited data obviated identification of any important subgroup that might benefit from higher doses. Implications for research Larger clinical trials with definitive outcomes are necessary to understand whether doses of RIF up to 40 mg/kg could safely improve treatment outcomes or reduce the duration of first-line therapy. The results of ongoing and planned RCTs of both pulmonary (for example, TRUNCATE-TB, RIFASHORT) and disseminated (NEW-STRAT-TB) TB will increase the certainty of the evidence for the higher doses tested in the studies in this review. Trials of higher doses of RIF in tuberculous meningitis should also be conducted. Now that the maximum tolerated dose of RIF in humans has been determined (40 mg/kg), studies of drug–drug interactions at this dose level will be essential to understand the effectiveness of antiretroviral therapy with higher RIF doses and to facilitate enrolment of more people with TB–HIV co-infection into future trials at higher doses. Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 286 3.8 Acknowledgements We acknowledge the WHO Global TB Programme, which commissioned and funded this review and provided valuable guidance and comments during its preparation. 3.9 References 1. Global tuberculosis report 2020. Geneva: World Health Organization; 2020 (https://www.who.int/ teams/global-tuberculosis-programme/tb-reports/9789240013131). 2. Espinal MA, Kim SJ, Suarez PG, Kam KM, Khomenko AG, Migliori GB et al. Standard short- course chemotherapy for drug-resistant tuberculosis: treatment outcomes in 6 countries. JAMA. 2000;283(19):2537–45. 3. Fox W, Ellard GA, Mitchison DA. Studies on the treatment of tuberculosis undertaken by the British Medical Research Council tuberculosis units, 1946–1986, with relevant subsequent publications. Int J Tuberc Lung Dis. 1999;3(10 Suppl 2):S231–79. 4. Short-course chemotherapy in pulmonary tuberculosis. A controlled trial by the British Thoracic and Tuberculosis Association. Lancet. 1976;ii(7995):1102–4. 5. Khan FA, Minion J, Al-Motairi A, Benedetti A, Harries AD, Menzies AD. An updated systematic review and meta-analysis on the treatment of active tuberculosis in patients with HIV infection. Clin Infect Dis. 2012;55(8):154–1163. 6. Guidelines for treatment of drug-susceptible tuberculosis and patient care (2017 update) (WHO/HTM/ TB/2017.05). Geneva: World Health Organization; 2017 (https://www.who.int/tb/publications/2017/ dstb_guidance_2017/en/). 7. Lienhardt C, Nunn A, Chaisson R, Vernon AA, Zignol M, Nahid P et al. Advances in clinical trial design: Weaving tomorrow’s TB treatments. PLoS Med. 2020;17(2):e1003059. 8. Jayaram R, Gaonkar S, Kaur P, Suresh BL, Mahesh BN, Jayashree R et al. Pharmacokinetics- pharmacodynamics of rifampin in an aerosol infection model of tuberculosis. Antimicrob Agents Chemother. 2003;47(7):2118–24. 9. Boeree MJ, Diacon AH, Dawson R, Narunsky K, du Bois J, Venter A et al. A dose-ranging trial to optimize the dose of rifampin in the treatment of tuberculosis. Am J Respir Crit Care Med. 2015;191(9):1058–65. 10. Boeree MJ, Heinrich N, Aarnoutse R, Diacon AH, Dawson R, Rehal S et al. High-dose rifampicin, moxifloxacin, and SQ109 for treating tuberculosis: a multi-arm, multi-stage randomised controlled trial. Lancet Infect Dis. 2017;17(1):39–49. 11. Velásquez GE, Brooks MB, Coit JM, Pertinez H, Vargas Vásquez D, Sánchez Garavito E et al. Efficacy and safety of high-dose rifampin in pulmonary tuberculosis. A randomized controlled trial. Am J Respir Crit Care Med. 2018;198(5):657–66. 12. Ruslami R, Nijland HM, Alisjahbana B, Parwati I, van Crevel R, Aarnoutse RE. ‘Pharmacokinetics and tolerability of a higher rifampin dose versus the standard dose in pulmonary tuberculosis patients. Antimicrob Agents Chemother. 2007;51(7):2546–51. 13. Heemskerk AD, Bang ND, Mai NT, Chau TT, Phu NH, Loc PP et al. Intensified antituberculosis therapy in adults with tuberculous meningitis. N Engl J Med. 2016;374(2):124–34. 14. Lienhardt C, Cook SV, Burgos M, Yorke-Edwards V, Rigouts L, Anyo G et al. Efficacy and safety of a 4-drug fixed-dose combination regimen compared with separate drugs for treatment of pulmonary tuberculosis: the Study C randomized controlled trial. JAMA. 2011;305(14):1415–23. 15. Technical report on the pharmacokinetics and pharmacodynamics (PK/PD) of medicines used in the treatment of drug-resistant tuberculosis (WHO/CDS/TB/2018.6). Geneva: World Health Organization; 2018 (https://apps.who.int/iris/handle/10665/260440). Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2 87 16. Hu Y, Liu A, Ortega-Muro F, Alameda-Martin L, Mitchison D, Coates A. High-dose rifampicin kills persisters, shortens treatment duration, and reduces relapse rate in vitro and in vivo. Front Microbiol. 2015;6:641. 17. Prideaux B, Via LE, Zimmerman MD, Eum S, Sarathy J, O’Brien P et al. The association between sterilizing activity and drug distribution into tuberculosis lesions. Nat Med. 2015;21(10):1223–7. 18. Mitchison DA. Basic mechanisms of chemotherapy. Chest. 1979;76(6 Suppl):771–81. 19. Imperial MZ, Nahid P, Phillips PPJ, Davies GR, Fielding K, Hanna D et al. A patient-level pooled analysis of treatment-shortening regimens for drug-susceptible pulmonary tuberculosis. Nat Med. 2018;24(11):1708–15. 20. Stott KE, Pertinez H, Sturkenboom MGG, Boeree MJ, Aarnoutse R, Ramachandran G et al. Pharmacokinetics of rifampicin in adult TB patients and healthy volunteers: a systematic review and meta-analysis. J Antimicrob Chemother. 2018;73(9):2305–13. 21. Ding J, Thuy Thuong Thuong N, Pham TV, Heemskerk D, Pouplin T, Tran CTH et al. Pharmacokinetics and pharmacodynamics of intensive antituberculosis treatment of tuberculous meningitis. Clin Pharmacol Ther. 2020;107(4):1023–33. 22. Moher D Liberati A, Tetzlaff J, Altman DG, the PRISMA Group Preferred Reporting Items for Systematic Reviews and Meta-Analyses. The PRISMA Statement. PLoS Med. 2009;6(7):e1000097 (doi. org/10.1371/journal.pmed.1000097). 23. Moher D, Shamseer L, Clarke M, Ghersi D, Liberati A, Petticrew M et al. Preferred reporting items for systematic review and meta-analysis protocols (PRISMA-P) 2015 statement. Syst Rev. 2015;4(1):1. 24. Definitions and reporting framework for tuberculosis – 2013 revision (WHO/HTM/TB/2013.2). Geneva: World Health Organization; 2013 (https://apps.who.int/iris/bitstream/handle/10665/79199/9789241505345_ eng.pdf). 25. Lefebvre C, Glanville J, Briscoe S, Littlewood A, Marshall C, Metzendorf MI et al. Technical supplement to Chapter 4: Searching for and selecting studies. In: Higgins J, Thomas J, Chandler J, Cumpston M, Li T, Page M et al., editors. Cochrane handbook for systematic reviews of interventions, version 6. Oxford: Cochrane Collaboration; 2019. 25. Bonnett LJ, Ken-Dror G, Koh G, Davies GR. Comparing the efficacy of drug regimens for pulmonary tuberculosis: Meta-analysis of endpoints in early-phase clinical trials. Clin Infect Dis. 20i7;65(1):46–54. 26. Covidence systematic review software 2020. Melbourne: Veritas Health Innovation; 2020 (www. covidence.org). 27. R Core Team. R: A language and environment for statistical computing. Vienna: R Foundation for Statistical Computing; 2018 (https://www.R-project.org/). 28. Higgins JP, Altman DG, Gøtzsche PC, Jüni P, Moher D, Oxman AD et al. The Cochrane Collaboration’s tool for assessing risk of bias in randomised trials. BMJ. 2011;43:D5928. 29. Jüni P, Altman DG, Egger M. Systematic reviews in health care: Assessing the quality of controlled clinical trials. BMJ. 2001;323(7303):42–6. 30. DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials. 1986;7(3):177–88. 31. Guyatt GH, Oxman AD, Vist GE, Kunz R, Falck-Ytter Y, Alonso-Coello P et al. GRADE: an emerging consensus on rating quality of evidence and strength of recommendations. BMJ. 2008;336(7650):924–6. 32. Guyatt GH, Oxman AD, Kunz R, Brozeka J, Alonso-Coello P, Rind D et al. GRADE guidelines. 6. Rating the quality of evidence – imprecision. J Clin Epidemiol. 2011;64:1283–93. 33. Merle C, Floyd S, Ndiaye A, Galperine T, Furco A, De Jong B et al. High-dose rifampicin tuberculosis treatment regimen to reduce 12-month mortality of TB/HIV co-infected patients: the RAFA trial results. In: 21st International AIDS Conference, 18-22 July 2016, Durban, South Africa Abstract WEAB0205LB available at https://www.abstract-archive.org/ 34. Long MW, Snider DE Jr, Farer LS. US Public Health Service Cooperative trial of three rifampin-isoniazid regimens in treatment of pulmonary tuberculosis. Am Rev Respir Dis. 1979;119(6):879–94. Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 288 35. Ruslami R, Ganiem AR, Dian S, Apriani L, Achmad TH, van der Ven AJ et al. Intensified regimen containing rifampicin and moxifloxacin for tuberculous meningitis: an open-label, randomised controlled phase 2 trial. Lancet Infect Dis. 2013;13(1):27–35. 36. Dian S, Yunivita V, Ganiem AR, Pramaesya T, Chaidir L, Wahyudi K et al. Double-blind, randomized, placebo-controlled phase II dose-finding study to evaluate high-dose rifampin for tuberculous meningitis. Antimicrob Agents Chemother. 2018;62(12):e01014-18. 37. Maug AKJ, Hossain MA, Gumusboga M, Decroo T, Mulders W, Braet S et al. First-line tuberculosis treatment with double-dose rifampicin is well tolerated. Int J Tuberc Lung Dis. 2020;24(5):499–505. 38. Yunivita V, Dian S, Ganiem AR, Hayati E, Hanggono Achmad T, Purnama Dewi A et al. Pharmacokinetics and safety/tolerability of higher oral and intravenous doses of rifampicin in adult tuberculous meningitis patients. Int J Antimicrob Agents. 2016;48(4):415–21. 39. Atwine D, Baudin E, Gelé T, Muyindike W, Mworozi K, Kyohairwe R et al. Effect of high-dose rifampicin on efavirenz pharmacokinetics: drug-drug interaction randomized trial. J Antimicrob Chemother. 2020;75(5):1250–8. 40. Heemskerk AD, Bang ND, Mai NT, Chau TT, Phu NH, Loc PP et al. Intensified antituberculosis therapy in adults with tuberculous meningitis. N Engl J Med. 2016;374(2):12–34. 41. Aarnoutse RE, Kibiki GS, Reither K, Semvua HH, Haraka F, Mtabho CM et al. Pharmacokinetics, tolerability, and bacteriological response of rifampin administered at 600, 900, and 1,200 milligrams daily in patients with pulmonary tuberculosis. Antimicrob Agents Chemother. 2017;61(11):e01054-17. 42. Jindani A, Borgulya G, de Patiño IW, Gonzales T, de Fernandes RA, Shrestha B et al. A randomised phase II trial to evaluate the toxicity of high-dose rifampicin to treat pulmonary tuberculosis. Int J Tuberc Lung Dis. 2016;20(6):832–8. 43. Higgins J, Thomas J, Chandler J, Cumpston M, Li T, Page M et al. Cochrane handbook for systematic reviews of interventions (v6.2). Oxford: Cochrane Collaboration; 2021 (https://training.cochrane.org/ handbook/current). 44. McGuinness L, Higgins J. Risk-of-bias VISualization (robvis): An R package and Shiny web app for visualizing risk-of-bias assessments. Res Synth Meth. 2020;12(1):55–61. 45. Te Brake LHM, de Jager V, Narunsky K, Vanker N, Svensson EM, Phillips PPJ et al. Increased bactericidal activity but dose-limiting intolerability at 50 mgkg(-1) rifampicin. Eur Respir J. 2021; 2000955 (doi: 10.1183/13993003.00955-2020. 46. Baker MA, Harries AD, Jeon CY, Hart JE, Kapur A, Lönnroth K et al. The impact of diabetes on tuberculosis treatment outcomes: a systematic review. BMC Med. 2011;9:81. 47. Steingart KR, Jotblad S, Robsky K, Deck D, Hopewell PC, Huang D et al. Higher-dose rifampin for the treatment of pulmonary tuberculosis: a systematic review. Int J Tuberc Lung Dis. 2011;15(3):305–16. 48. Onorato L, Gentile V, Russo A, Di Caprio G, Alessio L, Chiodini P et al. Standard versus high dose of rifampicin in the treatment of pulmonary tuberculosis: a systematic review and meta-analysis. Clin Microbiol Infect. 2021;27(6):830–7. 49. Mota L, Al-Efraij K, Campbell JR, Cook J, Marra F, Johnston J. Therapeutic drug monitoring in anti- tuberculosis treatment: a systematic review and meta-analysis. Int J Tuberc Lung Dis. 2016;20(6):81926. 50. Perumal R, Naidoo K, Naidoo A, Ramachandran G, Requena-Mendez A, Sekaggya-Wiltshire C et al. A systematic review and meta-analysis of first-line tuberculosis drug concentrations and treatment outcomes. Int J Tuberc Lung Dis. 2020;24(1):48–64. 51. Sileshi T, Tadesse E, Makonnen E, Aklillu E. The impact of first-line anti-tubercular drugs’ pharmacokinetics on treatment outcome: a systematic review. Clin Pharmacol. 2021;13:1–12. Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2 89 Appendix 1. Search strategy Search Set Central Medline Embase 1 Tuberculosis [tak] Tuberculosis [ms] Tuberculosis [mt] 2 Rifampicin [tak] OR Isoniazid [tak] OR Pyrazinamide [tak] OR Ethambutol [tak] Tuberculosis [tiab] Tuberculosis [tiab] 3 1 AND 2 1 OR 2 1 OR 2 4 Latent [tak] OR Prevention [tak] Rifampicin [tiab] OR Rifampin [tiab] OR Isoniazid [tiab] OR Pyrazinamide [tiab] OR Ethambutol [tiab] Rifampicin [tiab] OR Rifampin [tiab] OR Isoniazid [tiab] OR Pyrazinamide [tiab] OR Ethambutol [tiab] 5 3 NOT 4 Rifampin [sc] OR Isoniazid [sc] OR Pyrazinamide [sc] OR Ethambutol [sc] Rifampin [mt] OR Isoniazid [mt] OR Pyrazinamide [mt] OR Ethambutol [mt] 6 4 OR 5 4 OR 5 7 3 AND 6 3 AND 6 8 Randomized controlled trial [pt] OR Controlled clinical trial [pt] Crossover procedure [de] OR Double-blind procedure [de] OR Randomized controlled trial [de] OR Single-blind procedure [de] 9 Clinical trials as topic [mesh: no exp] Random* [deabti] OR Factorial* [deabti] OR Crossover* [deabti] OR Cross NEXT/1 over* [deabti] OR Placebo* [deabti] OR Doubl* NEAR/1 blind* [deabti] OR Singl* NEAR/1 blind* [deabti] OR Assign* [deabti] OR Allocat* [deabti] OR Volunteer* [deabti] 10 Randomized [tiab] OR Randomly [tiab] OR placebo [tiab] OR trial [ti] 8 OR 9 11 8 OR 9 OR 10 Animal experiment [mt] NOT (Human experiment [mt] OR Human [mt]) 12 Animals [mh] NOT Humans [mh] 10 NOT 11 13 11 NOT 12 7 AND 12 Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 290 Search Set Central Medline Embase 14 7 AND 13 Latent [tiab] OR Prevention [tiab] 15 Latent [tiab] OR Prevention [tiab] 13 NOT 14 16 14 NOT 15 [tak] word in title abstract or keyword, [pt] Publication Type, [ti ] word in title, [tiab] word in title or abstract, [sh] subheading, [mh] exploded MeSH term, [mesh: noexp] unexploded MeSH term, [mt] exploded Emtree term, [de] design, [deabti] Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2 91 Appendix 2. Characteristics of the studies included Aarnoutse et al. 2017 (1) Methods Double-blind RCT of 600, 900 or 1200 mg/day RIF for treatment of pulmonary TB for the first 2/12 of therapy with standard therapy with isoniazid, pyrazinamide and ethambutol. 600 mg documented as equivalent to 10 mg/kg in the study population; thus, 900 mg ≈ 15 mg/kg and 1200 mg ≈ 20 mg/kg. DOT for 2/12 treatments. Intensive sampling for PK in subset of participants. Follow-up: to 12/52 Loss to follow-up: 0 Participants Setting: Inpatients and outpatients (n=43, 29%) at several hospitals (Kilimanjaro Clinical Research Institute, Kilimanjaro Christian Medicine Centre, Kibong’oto National TB Hospital, Mawenzi Regional Hospital, Ifakara Health Institute) in the United Republic of Tanzania Number of participants: 150; 50 initially randomized to each of the three arms. 5 excluded from bacteriological analysis owing to absence of positive culture on LJ or MGIT (n=4) or isoniazid resistance at baseline (n=1). Bacteriological analysis results available for 49 patients at 600 mg, 48 at 900 mg and 48 at 1200 mg PK data available for 23 patients at 600 mg, 21 900 mg and 19 at 1200 mg Inclusion criteria: – Age 18–65 years – New, ZN smear-positive pulmonary TB (confirmed with AccuProbe assay or MGIT) – Female patients agreed to take measure to prevent pregnancy during initial phase of treatment Exclusion criteria: – Treatment with anti-TB drugs in past 3 years – Body weight < 50 kg – ALT or AST activity > 3 times upper limit of normal – Clinical liver disease (presenting with nausea, jaundice or tender hepatomegaly) – Serum creatinine above the upper limit of normal – Relevant history or current condition that might interfere with drug absorption, distribution, metabolism or excretion – Use of ART or expected use of ART within 2/12 Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 292 – Karnofsky score < 40 – Pregnant or breast-feeding – Rifampicin-resistant or MDR-TB HIV status: n=15 (10%) HIV positive; however unclear if all tested Baseline drug susceptibility: all tested for resistance to streptomycin, isoniazid, RIF and ethambutol at screening Intervention 600 mg, 900 mg or 1200 mg RIF daily from start of treatment for intensive phase, then switch to normal dose of RIF for continuation phase Study outcomes Bacteriological outcomes: – Time to culture conversion in LJ and in MGIT – Proportion of patients who achieved culture conversion after 4, 6, 8, 10 and 12/52 – Change in time to positivity in MGIT and number of log cfu on Middlebrook 7H11 plates over time PK outcomes: – Geometric mean total exposure to RIF – Safety/tolerability of higher doses of RIF Primary outcomes of interest Treatment success Definition: culture conversion on MGIT or LJ at 12/52 MGIT 33/49 (67%) at 600 mg), 26/48 (54%) at 900 mg, 32/48 (67%) at 1200 mg LJ 42/49 (86%) at 600 mg, 37/48 (77%) at 900 mg, 42/48 (88%) at 1200 mg Treatment failure Not defined, but no culture conversion on MGIT or LJ at 12/52 MGIT 16/49 (33%) at 600 mg, 22/48 (46%) at 900 mg, 16/48 (33%) at 1200 mg LJ 7/49 (14%) at 600 mg, 11/48 (23%) at 900 mg, 6/48 (13%) at 1200 mg Relapse No data. Follow-up period would not be long enough to capture. Death 3: 1 per group (1/50), i.e., 2% per group None considered related to RIF Adverse events 1004 adverse events: 821 grade 1, 160 grade 2, 20 grade 3 and 3 grade 5 (deaths) Secondary outcomes of interest Serious adverse events 7 at 600 mg, 6 at 900 mg and 10 at 1200 mg group Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2 93 6 events occurred in 2 patients at 600 mg (with, in total, 5 of the 6 events at least possibly related to RIF in both patients); 5 events (with 1 of the 5 events possibly related to RIF) occurred in 4 patients at 900 mg; and 9 adverse events occurred in 5 patients at 1200 mg (with 5 of the 9 events at least possibly related to RIF in 3 patients) Discontinuation Not documented Drug-specific adverse events: DILI, rifamycin hypersensitivity syndromes Grade 3 increases in AST/ALT: 1/50 (2%) 600 mg, 0/50 (0%) 900 mg, 2/50 (4%) 1200 mg Thrombocytopenia (“decreased platelet count”, scale not documented): 0/50 (0%) 600 mg, 2/50 (4%) 900 mg, 0/50 (0%) 1200 mg Hypersensitivity: 0 in all arms Disease-specific efficacy outcomes of interest: extrapulmonary TB Not available Notes Authors: most based in the Netherlands, some at recruitment sites and collaborators from Germany and the United Kingdom. On behalf of PanACEA consortium. Funding: NACCAP, a Dutch contribution to the European and Developing Countries Clinical Trials Partnership (EDCTP) that enabled the African Poverty Related Infection Oriented Research Initiative (APRIORI) and by the European and Developing Countries Clinical Trials Partnership (EDCTP) that funded the Pan-African Consortium for the Evaluation of Antituberculosis Antibiotics (PanACEA Consortium; project code IP.2007.32011.012 [HIGHRIF]) and Rapid Evaluation of Moxifloxacin in Tuberculosis (REMoxTB for microbiology support; project code IP.2007.32011.011) Atwine et al. 2020 (2) Methods Phase 2, open-label, 3-arm, drug–drug interaction, parallel, RCT in a 1:1:1 randomization ratio of patients with newly diagnosed HIV-associated pulmonary TB Groups given RIF 10 mg/kg and HZE with EFV 600 mg, RIF 20 mg/kg and HZE with EFV 600 mg, RIF 20 mg/kg and HZE with EFV 800 mg orally for 8/52 (intensive phase) then back to standard dosing for continuation phase ART started 2–4/52 after starting TB treatment Recruited March 2014–August 2016 Follow-up: to 28/52 Lost to follow-up: 2: 1/33 (3%) R10EFV600, 1/31 (3%) R20EFV600, 0/33 (3%) R20EFV800 Participants Setting: Mbarara, south-west Uganda; not otherwise described. Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 294 Number of participants: 98 randomized – 33 to R10EFV600, 32 to R20EFV600, 33 to R20EFV800 One at R20EFV600 found to be HIV negative and excluded Inclusion criteria: – adults – RIF-susceptible pulmonary TB confirmed by Xpert MTB/RIF – HIV positive, ART-naïve, CD4 count 50–250 cells/mm3 – body weight > 45 kg – “without medical contraindications” – protocol later amended to include body weight > 35 kg and removal of CD4 cell count limit Exclusion criteria: – RIF resistance on Xpert MTB/RIF and confirmed by Genotype MTBDRplus assay – concomitant opportunistic infection requiring additional infection medication – Karnofsky score < 80 – ALT or bilirubin > 5x upper limit of normal (hepatitis grade 3 or 4) – haemoglobin < 7.5 g/dL (grade 3 or 4) – grade 4 clinical sign or biological result according to the Division of AIDS Table for Grading the Severity of Adult and Paediatric Adverse Events (version 2.0, Nov 2014) – patient unable to give informed consent or unlikely to cooperate with sampling procedures – patient with psychiatric illness that might prevent follow-up according to protocol – patient receiving or requiring medications that might interfere with study drugs HIV status: all tested as part of inclusion criteria. 97/98 of those randomized (other patients excluded as HIV negative) Baseline drug susceptibility: tested for RIF resistance at screening Intervention RIF started immediately as part of TB treatment for first 2/52. All had standard dose RIF for continuation phase EFV started at 2-4/52 after commencing TB treatment Study outcomes Primary: – EFV PK parameters (Cmin, Cmax, Tmax and AUC0-24 Secondary: – RIF Cmax – genetic polymorphism of enzymes involved in EFV metabolism – ART and TB therapy efficacy end-points – treatment adherence – safety end-points Primary outcomes of interest Treatment success Definition: “cured and completed”. 8/52 sputum culture conversion data for LJ and MGIT given 29/33 (88%) R10EFV600, 28/31 (90%) R20EFV600, 31/33 (94%) R20EFV800 Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2 95 Treatment failure 1/33 R10EFV600 (patient found to have MDR-TB in bone marrow by Xpert) 0 in both R20 groups Relapse No data. Follow-up period would not be sufficient to capture Death 3 in total, 1 per group: 1/33 (3%) R10EFV600, 1/31 (3%) R20EFV600, 1/33 (3%) R20EFV800 None considered to be related to RIF Adverse events Total grade 3 or 4 adverse events: 23 in total R10EFV600, 14 in R20EFV600, 16 in R20EFV800 Secondary outcomes of interest Serious adverse events 6/33 (18%) R10EFV600, 6/31 (19%) R20EFV600, 6/33 (18%) R20EFV800 Discontinuation 3 in total, 1 per group: 1/33 (3%) R10EFV600, 1/31 (3%) R20EFV600, 1/33 (3%) R20EFV800 Drug-specific adverse events: DILI, rifamycin hypersensitivity syndromes Grade 3 or above increase in transaminase activity: 2/33 (6%) R10EFV600, 2/32 (6%) R20EFV600, 2/33 (6%) R20EFV800 Thrombocytopenia (scale not specified): 4/33 (12%) R10EFV600, 2/32 (6%) R20EFV600, 3/33 (9%) R20EFV800 Disease-specific efficacy outcomes of interest, i.e., extrapulmonary TB NA Notes Authors: based at study site in Uganda and in France Funding: Agence nationale de recherches sur le sida et les hépatites virales (ANRS), Paris, France, as the sponsor (grant number ANRS12292), and funded jointly with Médecins Sans Frontières Boeree et al., 2017 (3) Methods RCT, open label, multi-arm, multi-stage design of patients with pulmonary TB. Randomly assigned in 1:1:1:1:2 ratio to receive: – 35 mg/kg RIF with 15–20 mg/kg ethambutol – 20 mg/kg RIF with 400 mg moxifloxacin – 20 mg/kg RIF with 300 mg SQ109 – 10 mg/kg RIF with 300 mg SQ109 – control regimen; 10 mg/kg RIF, 5 mg/kg H, 25 mg/kg Z, 15–20 mg/kg E Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 296 Experimental treatments also had H and Z at normal doses. Follow-up: 12/12 follow-up in total, but ITT analysis done at 6/12 for culture conversion. Seen once a week up to week 12, then at weeks 14, 17, 22 and 26 after start of treatment Lost to follow-up: ITT analysis done at 6/12, when LTFU 6/123 RHZE, 4/63 R35HZE, 6/59 R10QHZ, 3/57 R20HZQ, 3/63 R20HZM Participants Setting: Three clinical trial sites in the United Republic of Tanzania and four in South Africa, including hospitals, health centres and clinical trial units Number of participants: 365; 63 to R35HZE, 57 to R20HZQ, 63 to R20HZM, 59 to R10HZQ, 123 HRZE Recruitment stopped early in SQ109 arms as pre-specified efficacy thresholds were not met in planned interim analysis Inclusion criteria: – age ≥ 18 years – weight 35–90 kg – newly diagnosed, previously untreated pulmonary TB – confirmed to be RIF sensitive by Xpert MTB/RIF – positive smear microscopy of at least 1+ on IUATLD/WHO scale – patients with HIV were eligible if CD4 > 200 cells/mm3 and if local ethics committee agreed that ART could be safely withheld until study week 12 Exclusion criteria: – pregnant or breastfeeding – receiving therapy expected to prolong QT interval on electrocardiogram or alter cytochrome P450 enzyme activity with potential effects on SQ109 metabolism HIV status: all tested Control 9/123, R35HZE 4/63, R20HZQ 3/57, R20HZM 3/63, R10HZQ 5/59 Baseline drug susceptibility: phenotypic drug resistance checked for HRZE and moxifloxacin: 2/365 resistant to RIF, 8/365 resistant to H, 12/365 resistant to Z, 1/365 resistant to E, 0/365 resistant to moxifloxacin Intervention Started immediately on commencing TB treatment Study outcomes Primary end-point: – time from treatment initiation to the first of two consecutive negative once-weekly sputum cultures without an intervening positive culture in liquid media, up to 12/52 Secondary end-points: – time to first negative culture in liquid and solid media – proportion of patients who converted to negative sputum culture in liquid and solid media at each time Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2 97 – rate of change in time to positivity in liquid culture – frequency of treatment failure in each treatment arm – development of drug resistance in experimental arms – safety data Primary outcomes of interest Treatment success Definition: two consecutive negative once-weekly sputum cultures without an intervening positive culture in liquid media, up to 12/52 Liquid culture data (ITT): RIF35HZE 51/63 (81%), RIFQHZ 44/59 (75%), RIF20QHZ 48/57 (84%), RIF20MHZ 52/63 (83%), RHZE 101/123 (82%) PP data also available Treatment failure 11/365 in total Relapse Data provided for recurrence of disease. Data on whether clear reinfection or relapse not yet published Recurrence (with culture confirmation): 2/123 RHZE, 0/63 R35HZE, 1/59 R10HZQ, 3/57 R20HZQ, 0/63 R20HZM Death At 6/12: 1/63 R35HZE only Adverse events 285/365 patients had at least 1 adverse event 45/365 patients had at least 1 grade 3, 4 or 5 adverse event Secondary outcomes of interest SAEs 23/365 patients had at least one SAE Discontinuation Treatment changed due to hepatic adverse event 10/365 2/123 RHZE, 3/63 RIF35HZE, 0/59 R10HZQ, 3/57 RIF20HZQ, 0/63 R20HZM Drug-specific adverse events: DILI, rifamycin hypersensitivity syndromes Hepatotoxicity (grade 3 or above): 3/63 R35HZE, 1/59 R10HZQ, 1/57 R20HZQ, 0/63 R20HZM, 1/123 RHZE No report of hypersensitivity syndromes Disease-specific efficacy outcomes of interest, i.e. extrapulmonary TB Not available Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 298 Notes Authors: In Germany and the Netherlands, with some site collaborators in South Africa and the United Republic of Tanzania On behalf of the PanACEA consortium. Funding: European and Developing Countries Clinical Trials partnership (EDCTP), the German Ministry for Education and Research (BmBF) and the Medical Research Council UK (MRC) Dian et al., 2018 (4) Methods Double-blinded, randomized, placebo-controlled phase II trial with three parallel arms in patients with tuberculous meningitis. Not necessarily first episode of TB. Randomly assigned in 1:1:1 to oral: – 10 mg/kg RIF and HZE – 20 mg/kg RIF and HZE – 30 mg/kg RIF and HZE for 1/12, before completing standard dose RHZE to 2/12 and subsequent RH at normal dosing for at least 4/12. Recruited December 2014–November 2016 Follow-up: until 6/12 after treatment started Lost to follow-up: 0, all followed up to 6/12 Participants Setting: Hasan Sadikin Hospital, Bandung, Indonesia (referral hospital for West Java) Number of participants: 60 (20 in each group) Inclusion criteria: – age ≥ 15 years – clinical suspicion of tuberculous meningitis and CSF/blood glucose ratio < 0.5 – none or < 3/7 anti-TB chemotherapy taken for current infection – participants who could become pregnant agree to use at least one form of non-hormonal contraception from date of informed consent through to end of study – willing to allow storage of specimens – patient or representative (if patient incapacitated) willing and able to give informed consent Exclusion criteria: – no CSF data available for microbiological confirmation (definite) or probable tuberculous meningitis – liver dysfunction (ALT > 5x upper limit of normal), kidney dysfunction (estimated glomerular filtration rate < 50 mL/min) – pregnant or breastfeeding – confirmed cryptococcal meningitis or diagnosis of bacterial meningitis based on clinical assessment and routine CSF examination Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2 99 – rapid clinical deterioration at time of presentation (considered unethical to conduct a study in these patients with PK measures as primary end-point) – history of hypersensitivity or intolerance to RIF – potential non-compliant subjects as judged by the investigator HIV status: all tested; 1/20 RIF10, 1/20 RIF20, 4/20 RIF30 Baseline drug susceptibility: unclear, RIF resistance not an exclusion criterion (in view of the time required for resistance testing in a resource-poor setting), but one patient was withdrawn from the study for RIF resistance Intervention RIF started at commencement of TB treatment for 1/12, then participants continued standard treatment Placebo tablets, participants had same number of tablets Study outcomes Primarily PK outcomes: – demonstrate any difference in exposure to RIF (AUC0-24) between groups Secondary outcomes: – safety and tolerability of regimens – mortality (not enough power to detect difference) Primary outcomes of interest Treatment success Definition: not defined Treatment failure Not defined Relapse Not documented Death Data at 6/12 (also available at discharge 30, 45 and 60 days) Overall mortality 19/60 (32%): 10 mg/kg 7/20, 20 mg/kg 9/20, 30 mg/kg 3/20 Adverse events Total number grades 1–2 adverse events (not number of participants) 51 Total number grades 3-–4 adverse events 15 Secondary outcomes of interest Serious adverse events Not specified Discontinuation Not specified Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2100 Drug-specific adverse events: DILI, rifamycin hypersensitivity syndromes Hepatotoxicity grades 3 and 4 (number of events): 12: 10 mg/kg 3/30, 20 mg/kg 5/20, 30 mg/kg 4/20 Thrombocytopenia grades 1 and 2 only: 8: 10 mg/kg 2/20, 20 mg/kg 4/20, 30 mg/kg 2/20 No hypersensitivity reactions reported Disease-specific efficacy outcomes of interest i.e. extrapulmonary TB New neurological events not reported as adverse events as end-point to assess adverse events possibly/probably related to treatment. Reported, however, that new neurological events as well as functional outcome were equally distributed among the three groups of patients (data not shown). Notes Authors: in the Netherlands and Indonesia Funding: Ministry of Research, Technology and Higher Education, Indonesia Heemskerk et al. 2016 (5) Methods Randomized, double-blind, placebo-controlled trial of patients with clinical diagnosis tuberculous meningitis. Randomized to: – standard 9/12 tuberculous meningitis treatment, i.e., 10 mg/kg RIF – intensified regimen of 15 mg/kg RIF and levofloxacin with HZE Follow-up: 9/12; until end of treatment Lost to follow-up: 53: 28/409 on standard treatment, 25/408 on intensified treatment Participants Setting: Pham Ngoc Thach Hospital for Tuberculosis and Lung Disease and the Hospital for Tropical Diseases, both in Ho Chi Minh City, Viet Nam Number of participants: 817: 409 on standard treatment plus placebo, 408 on intensified treatment Inclusion criteria: – age ≥ 18 years – clinical diagnosis of tuberculous meningitis (at least 5 days of meningitis symptoms, nuchal rigidity and CSF abnormalities) Exclusion criteria: – positive CSF Gram or India Ink stain – pregnant – known hypersensitivity or intolerance to fluoroquinolones or rifamycin – creatinine > 3x upper limit of normal – bilirubin > 2.5x upper limit of normal, AST or ALT > 5x upper limit of normal Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2 101 – MDR-tuberculous meningitis (on the basis of previous sputum drug susceptibility results or Xpert MTB/RIF) – Llack of informed consent HIV status: all tested. 349/817 HIV positive. 174/409 (42.5%) standard treatment, 175/408 (42.9%) intensified treatment Baseline drug susceptibility: denominator values for participants for whom drug susceptibility results were available H monoresistance 41/156 (26.3%) standard, 113/166 (68.1%) intensified R monoresistance 1/156 (0.6%) standard, 0/166 intensified MDR 7/156 (4.5%) standard, 8/166 (4.8%) intensified Intervention Treatment started as soon as participant was randomized Placebo tablets with same appearance given to patients in standard treatment group Study outcomes Primary outcome: – death by 9/12 after randomization Secondary outcomes: – neurological disability at 9/12 – time to first new neurological event or death – SAEs Primary outcomes of interest Treatment success Definition: not defined Treatment failure Not defined Relapse Not documented Deaths 227/817: 114/409 (27.9%) standard ITT, 113/408 (27.7%) intensified ITT PP values also available Adverse events Number of participants experiencing event: 229/409 (56.0%) standard, 240/408 (58.8%) intensified Secondary outcomes of interest SAEs 58/409 (14.2%) standard, 75/408 (18.4%) intensified Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2102 Discontinuation Not documented Drug-specific adverse events: DILI, rifamycin hypersensitivity syndromes Hepatotoxicity: 28/409 (6.8%) standard, 17/408 (4.2%) intensified Thrombocytopenia: 17/409 (4.2%) standard, 14/408 (3.4%) intensified Disease-specific efficacy outcomes of interest, i.e., extrapulmonary TB Neurological sequelae: 144/409 (35.2%) standard, 148/408 (36.3%) intensified Notes Authors: in Oxford and Liverpool (UK) and Ho Chi Minh (Viet Nam) Funding: Wellcome Trust and Li Ka Shing Foundation Jindani et al., 2016 (42) Methods RCT of patients with newly diagnosed pulmonary TB. Only laboratory staff and independent clinician who made a clinical assessment of any SAE were blinded. Randomized 1:1:1 to: – standard treatment RHZE with 10 mg/kg RIF – RIF 15 mg/kg for 16/52 and HZE – RIF 20 mg/kg for 16/52 and HZE Follow-up: participants attended daily for 8/52 for DOT, then either attended 6 days a week or were given medications for domiciliary treatment monitoring. Participants were interviewed about symptoms and signs, and serum ALT activity was measured at 2, 4, 8, 12 and 16 weeks Lost to follow-up: none Participants Settings: Santa Cruz, Bolivia; Kathmandu, Nepal; Mbarara, Uganda Number of participants: 300; 100/arm: 150 in Bolivia, 50 in Nepal, 100 in Uganda Inclusion criteria: – aged 18–65 years – two sputum samples positive for tubercle bacilli on microscopy – received < 1/12 of previous TB therapy – reliable and accessible home address Exclusion criteria: – critically ill – extrapulmonary TB – alcoholism – pregnancy – psychiatric illness – blood disorders Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2 103 – diabetes – epilepsy – peripheral neuritis – haemoglobin < 7 g/dl – ALT > 5x upper limit of normal – creatinine clearance < 30mL/min – isoniazid or RIF resistance HIV status: all tested: All negative Baseline drug susceptibility: 4 patients had initial drug resistance. 10 mg/kg 1/100, 20 mg/kg 2/100, 30 mg/kg 1/100 Intervention Treatment started as soon as participant was randomized. Study outcomes Primary outcome: – any grade 3 or 4 adverse event and/or SAE during first 16/52 of chemotherapy Secondary outcomes: – culture conversion at the end of 8/52 of chemotherapy – treatment modification as a result of an SAE and/or grade 3 or more adverse event – any increase in ALT activity at any time during treatment – number of observed doses of chemotherapy ingested Primary outcomes of interest Treatment success Definition: authors reviewed culture conversion at end of 8/52 chemotherapy – single negative culture only 10 mg/kg 69/92 (75.0%), 15 mg/kg 66/80 (82.5%), 20 mg/kg 76/91 (83.5%) (per protocol) Treatment failure Not documented Relapse Not documented Death 2/300: 1 at 15 mg/kg, from oesophageal carcinoma; 1 at 20 mg/kg, from septicaemia Neither considered to be treatment related Adverse events 10 mg/kg 20/100, 15 mg/kg 22/100, 20 mg/kg 13/100 Secondary outcomes of interest SAEs 2/300: both deaths as above Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2104 Discontinuation No discontinuations 1 interruption at 15 mg/kg Drug-specific adverse events: DILI, rifamycin hypersensitivity syndromes Hepatoxicity grade 3 (no grade 4 events) number of events. 6 levels taken for each participant (Numbers represent number of tests, not number of participants): 10 mg/kg 1/600, 15 mg/kg 2/600, 20 mg/kg 4/600 Thrombocytopenia not documented Cutaneous hypersensitivity: 1/100 patient at 15 mg/kg Disease specific efficacy outcomes of interest i.e. extrapulmonary TB Not available Notes Authors: from St George’s Hospital, London; Bolivia, Nepal and Uganda, with collaborators from the USA and Epicentre MSF Funding: jointly by St George’s Hospital and University of London, UK, and Epicentre MSF Long et al. 1979 (7) Methods RCT of patients with pulmonary TB. Randomized to: – 450 mg RIF – 600 mg RIF – 750 mg RIF All with H for 20/52, then receive H and E for 12 or 18/12 after sputum cultures become negative Recruited October 1970–September 1972 Follow-up: length of follow-up differed by arm – minimum, 1 year Lost to follow-up: 43/822 (5.2%): R450 9/167 (5.3%), R600 17/324 (5.2%), R750 17/331 (5.1%) Participants Setting: 18 hospitals and 34 clinics throughout the USA Number of participants: 822: R450 167, R600 324, R750 331 Inclusion criteria: – age > 13 years – sputum containing acid-fast organisms – received no more than 14/7 anti-TB therapy – chest X-ray evidence of active pulmonary TB with or without cavitation Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2 105 Exclusion criteria: – pre-existing renal, hepatic, haematological or ophthalmological conditions that, in the opinion of the attending physician, introduced undue risks of adverse reactions to the drugs – patients on corticosteroids – patients who probably could not be observed for 3 years (terminally ill, temporary visa or migrant worker) HIV status: not documented Baseline drug susceptibility: 39/668 (5.8%) resistance to RIF, ethambutol or isoniazid; data only for patients who completed initial treatment 20 resistant to INH alone, 7 to RIF alone, 1 to EMB alone, 3 to RIF and INH, 6 to INH and EMB and 2 to all three Intervention Treatment started as soon as the participant was randomized Study outcomes Outcomes: – to compare the efficacy and toxicity of 3 different doses of RIF in combination with INH during initial therapy – to compare 12 months versus 18 months of chemotherapy after sputum conversion Primary outcomes of interest Treatment success Definition: bacteriological conversion of sputum at 20/52, data for 552 participants as others excluded for resistance, negative initial cultures, contaminated cultures, pretreatment cultures containing non-tuberculous Mycobacterim or previous TB treatment for more than 14 days At 12/52: R450 12% positive cultures, R600 7%, R750 6%. Numerators and denominators not clear: 20/52: R450 7.7%, R600 0.5%, R750 0.9% positive cultures At 20/52: 127 R450, 269 R600 and 272 R750 completed treatment. Therefore 117/127 R450, 268/269 R600, 270/272 R750 Treatment failure Defined as clinically worse (and death, excluded from these numbers) 5 treatment failures (denominator unclear): R450 3, R600 1, R750 1 Later in the paper, it is reported that 16 patients should be considered as treatment failures – data inconsistent Relapse R450 1/64, R600 2/151, R750 0/171 Death 26/822 all causes: R450 8/167, R600 11/324, R750 7/331 3/822 from TB: R450 1/167, R750 2/331 Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2106 Adverse events 27/822 Secondary outcomes of interest SAEs Not documented Discontinuation 154, denominator unclear Drug-specific adverse events: DILI, rifamycin hypersensitivity syndromes Hepatotoxicity: R450 6/167, R600 3/324, R750 7/331 (some inconsistency in definitions) Thrombocytopenia: R450 28/167, R600 17/324, R750 39/331 Hypersensitivity: 0/822 Disease-specific efficacy outcomes of interest, i.e., extrapulmonary TB Not available Notes Authors: All at US Centers for Disease Control and Prevention, USA Funding: US Public Health Service Maug et al., 2020 (8) Methods Open-label 1:1 parallel RCT of patients with new or retreated pulmonary TB Randomized to: – standard RHZE with 10 mg/kg RIF – 20 mg/kg RIF with standard HZE Recruited November 2014–September 2015 Follow-up: 12/12 after cure or treatment completion Lost to follow-up: 16. 8/343 control, 8/347 intervention Participants Setting: Eight large diagnostic and treatment centres run by the Damien Foundation in Bangladesh Number of participants: 701: 348 control, 353 intervention Inclusion criteria: – age ≥ 15 years – smear-positive pulmonary TB Exclusion criteria: – RIF resistance at diagnosis – requiring hospitalization owing to poor clinical condition Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2 107 – clinically active liver disease – known hepatitis B or C or HIV infection – pregnancy HIV status: not tested. Known HIV-positive excluded. Authors assumed rough prevalence of 0.1% from previous research in patients with TB in Bangladesh Intervention Treatment started as soon as patient was randomized Study outcomes Primary end-points: – efficacy – occurrence of unfavourable treatment outcomes (relapse, failure, death or loss to follow-up) against relapse-free treatment success – safety – occurrence of any SAE and grade 3-4 hepatoxicity Primary outcomes of interest Treatment success Definition: assessed clinically and by sputum acid-fast Bacillus smear and culture 12/12 after cure or treatment completion: 290/348 control, 306/353 intervention (ITT) Treatment failure Failure based on positive smear for acid-fast Bacillus but not necessarily positive culture: 8/348 control, 12/353 intervention Relapse Relapsed within 12 months: 1/348 control, 3/353 intervention Death 11/348 control, 5/353 intervention Adverse events Not documented Secondary outcomes of interest Serious adverse events 24/348 control, 13/353 intervention Drug-related SAEs: 15/348 control, 9/353 intervention Discontinuation 0 To stop medication if transaminases > 5x upper limit of normal, but it appears that all were reintroduced (no documentation to suggest otherwise) Drug-specific adverse events: DILI, rifamycin hypersensitivity syndromes Hepatoxicity: Grade 3 or 4 increase in transaminases: 7/348 control, 3/353 intervention Hepatitis/jaundice: 6/348 control, 6/353 intervention Disease specific efficacy outcomes of interest i.e. extrapulmonary TB Not available Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2108 Notes Authors: first author in Bangladesh, others in Belgium with one collaborator in France Funding: Damien Foundation Belgium, Brussels, and Institute of Tropical Medicine, Antwerp, Belgium Merle et al. 2016 (9) Methods Multicentre, open-label RCT with three parallel arms of patients with HIV-associated TB of any form. Randomized to: – A: RHZE (standard 10 mg/kg RIF) with ART initiation at 2/52 – B: RHZE (standard 10 mg/kg RIF) with ART initiation at 8/52 – C: 15 mg/kg RIF with standard HZE for first 2/12 and ART initiation at 8/52 Follow-up: until 18/12 post-randomization Clinical visit and 2 sputum samples taken every 2/52 during 2/12, every 1/12 until end of TB treatment and every 3/12 until end of follow-up Lost to follow-up: before 12/12: A 21/251, B 21/247, C 19/249 (PP denominators) Participants Setting: Cotonou and Porto-Novo, Benin, Conakry, Guinea, and Dakar, Senegal Number of participants: 778: A 262, B 258, C 258 Later exclusions to take n to 747: 11 excluded from A for MDR, not ART naïve, other 11 excluded from B for MDR, CD4 < 50 cells/mm3 9 excluded from C for MDR, no proof of TB, other Inclusion criteria: – age ≥ 18 years – bacteriologically confirmed TB – HIV positive – ART naïve – CD4 ≥ 50 cells/mm3 at enrolment Exclusion criteria: Not specified HIV status: 100% Baseline drug susceptibility: not specified Intervention Treatment started as soon as participant was randomized Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2 109 Study outcomes Primary outcome: 12/12 mortality Primary outcomes of interest Treatment success Definition not specified. These figures presumed, all but those with failure, recurrence and death: A 241/251, B 240/247, C 243/249 (ITT) Treatment failure A 10/251, B 7/247, C 6/249 Relapse Recurrence data given: A 8/251, B 5/247, C 4/269 Death During treatment A 10/251, B 26/247, C 20/269 Adverse events Only hepatotoxicity documented (see below) Secondary outcomes of interest SAEs Not documented Discontinuation Not documented Drug-specific adverse events: DILI, rifamycin hypersensitivity syndromes Hepatotoxicity, grade 3 and 4: A 1/251, B 0/247, C 3/269 No documentation of thrombocytopenia or hypersensitivity reactions Disease-specific efficacy outcomes of interest, i.e., extrapulmonary TB Form of TB diagnosed not documented Notes Authors: first WHO (Switzerland) and London School of Hygiene and Tropical Medicine (UK), other authors from Belgium, Benin, Guinea, Senegal, South Africa and UK Funding: European & Developing Countries Clinical Trials Partnership Ruslami et al., 2007 (10) Methods Double-blind randomised phase-II clinical trial in patients with newly diagnosed, untreated pulmonary TB. Randomized to – 450 mg, i.e., ~10 mg/kg RIF and standard HZE – 600 mg, i.e., ~13 mg/kg RIF and standard HZE Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2110 Follow-up: 6/12 Lost to follow-up: 2, one in each group Participants Setting: Urban outpatient TB clinic in Bandung, Indonesia Number of participants: 50, 25 in each arm Inclusion criteria: – newly diagnosed (clinical symptoms, chest X-ray, confirmed by microscopic detection of acid- fast Bacillus and untreated TB – aged ≥ 18 Exclusion criteria: – body weight < 33 kg – pregnant or breastfeeding – history of liver or kidney disease or any other disease that might affect the PK of TB drugs – previous treatment for TB HIV status: all tested: 1/47 (data available only for participants who provided all PK data), 1/24 standard dose Baseline drug susceptibility: all tested for RIF, isoniazid, ethambutol and streptomycin resistance: 2 resistant to H, 1 resistant to R, 1 MDR (Not clear in which groups these participants were) Intervention Commenced after randomization Study outcomes Primary outcome: Effect of higher dose of RIF in terms of PK and tolerability Primary outcomes of interest Treatment success Definition: an initially smear-positive patient who was smear-negative in the last month of treatment and on at least one previous occasion: 38/45 – only total treatment success, not by group Treatment failure Definition: a smear-positive patient who remained smear-positive at month 5 or after during treatment: 3/45 – only total failure, not by group Relapse Not reported Death 0 Adverse events Incidence given as percentages per group, no raw data available. No data on how many participants had adverse events Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2 111 Secondary outcomes of interest SAEs 4/49 for both groups Discontinuation Not reported Drug-specific adverse events: DILI, rifamycin hypersensitivity syndrome Hepatotoxicity (grade 3): 10 mg/kg 3/24, 13 mg/kg 1/23 No data provided on thrombocytopenia 0 hypersensitivity reactions Disease specific efficacy outcomes of interest, i.e., extrapulmonary TB Not available Notes Authors: Indonesia and the Netherlands Funding: Supported by a grant from PRIOR, a fellowship for Rovina Ruslami from the Netherlands Foundation for Tropical Research (and from Radboud University, Nijmegen, and a clinical fellowship for Reinout van Crevel from the Netherlands Organization for Health Research and Development Ruslami et al., 2013 (11) Methods Open-label, phase-II RCT of factorial design of patients with tuberculous meningitis. First randomized to: – RIF 10 mg/kg orally – RIF 13 mg/kg IV Then randomized to: – moxifloxacin 400 mg – moxifloxacin 800 mg – ethambutol 75 0mg Also, standard dose isoniazid, pyrazinamide and adjunctive corticosteroids for 14/7, then continue with standard RHZE. Follow-up: 6/12 after start of treatment In the first 2/52 of treatment, tuberculous meningitis and possible drug-related adverse events were monitored daily, with twice weekly electrocardiogram, full blood count and liver transaminases. More frequent or other investigations were done as clinically indicated. After discharge, patients were reviewed monthly Lost to follow-up: 0 Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2112 Participants Setting: Hasan Sidikin Hospital, Bandung, Indonesia Number of participants: 60 31/60 10 mg/kg RIF orally ≥ 12 no moxifloxacin, 10 moxifloxacin 400 mg, 9 moxifloxacin 800mg 29/60 13 mg/kg RIF IV ≥ 10 no moxifloxacin, 9 moxifloxacin 400 mg, 10 moxifloxacin 800 mg Inclusion criteria: – age > 14 years – definite, probable or possible tuberculous meningitis Exclusion criteria: – no diagnostic lumbar puncture – evidence of bacterial or cryptococcal meningitis – treatment for TB for > 7 days before admission – history of tuberculous meningitis – pregnant or breastfeeding – known contraindication to moxifloxacin – ALT > 5x upper limit of normal – hypersensitivity or intolerance to RIF or moxifloxacin – rapid clinical deterioration during screening – no informed consent HIV status: all tested, 7/60 positive: 4/31 10 mg/kg, 3/29 13 mg/kg Baseline drug susceptibility: culture-confirmed cases (29/60) tested for resistance to RIF, isoniazid and streptomycin: 2/29 resistant to streptomycin; no resistance to RIF or isoniazid Intervention At time of randomization Study outcomes Primary outcome: PK and safety or tolerability of intensified treatment regimens Secondary outcomes: compare neurological outcomes and mortality Primary outcomes of interest Treatment success Not documented, mortality mark of success used Treatment failure Not documented Relapse Not documented Death 30/60: 20/31 10 mg/kg, 10/29 13 mg/kg Adverse events 34/60: 17/31 10 mg/kg, 17/29 13 mg/kg Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2 113 Secondary outcomes of interest SAEs 12/60: 5/31 10 mg/kg, 7/29 13 mg/kg Discontinuation Unclear; stopped in patients with grade 4 toxicity, but table gives grades 3 and 4 together Drug-specific adverse events: DILI, rifamycin hypersensitivity syndromes Hepatotoxicity grade 3 and 4: 11/60: 5/31 10 mg/kg, 6/29 13 mg/kg No haematological adverse events, so thrombocytopenia 0% Hypersensitivity: 3/60, all at 13 mg/kg: 2/29 grade 1 or 2, 1/29 grade 3 or 4 Disease-specific efficacy outcomes of interest, i.e. ,extrapulmonary TB Resolution of coma more rapid at 13 mg/kg: group median 4 days (IQR 2–8) vs 5 days (3–7). Complete neurological recovery at 6/12 of treatment: 13 mg/kg 9/29 vs 10 mg/kg 4/31 Notes Authors: in Indonesia and the Netherlands Funding: Royal Dutch Academy of Arts and Sciences, Netherlands Foundation for Scientific Research and Padjadjaran University, Bandung, Indonesia Velasquez et al., 2018 (12) Methods Blinded, randomized controlled phase-II trial of patients with new pulmonary TB. Randomized 1:1:1 to: – 10 mg/kg RIF and standard HZE – 15 mg/kg RIF and standard HZE – 20 mg/kg RIF and standard HZE for the first 2/12 and then standard continuation phase RH. Follow-up: to 12/12 after start of treatment Lost to follow-up: 4/180 Participants Setting: from 43 health centres referred to Hospital Nacional Hipolito Unanue or Hospital Sergio Bernales in Lima, Peru Number of participants: 180: 60 per arm Inclusion criteria: – age 18–60 years – previous TB treatment for > 1/12 – strains susceptible to RIF and isoniazid Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2114 Exclusion criteria: – weight < 30 kg – extrapulmonary manifestations (CNS, miliary, pericardial or pleural disease) – significant haemoptysis – study drug intolerance, hypersensitivity or contraindication – pregnancy or breastfeeding – ALT > 2x upper limit of normal – bilirubin > 2.5x upper limit of normal – creatinine > 2x upper limit of normal or clearance <60ml/min – haemoglobin < 7g/dL – platelets < 150 000 mm3 – white cell count < 4500 cells/µL – hepatitis B surface antigen or hepatitis C antibody seropositive – glycated haemoglobin > 7.5% – Karnofsky score < 50 HIV status: 5/180: 2/60 10 mg/kg, 2/60 15 mg/kg, 1/60 20 mg/kg Baseline drug susceptibilities: 0% resistance Intervention Started at randomization Study outcomes Primary outcomes: – change in elimination rate of M. tuberculosis log10 colony-forming units – Frequency of grade 2 or higher RIF-related adverse events Secondary outcomes: – Proportion of culture conversion on LJ media at 8/52 – Proportion of unfavourable outcomes (treatment failure, recurrence after cure, relapse, reinfection or death) Primary outcomes of interest Treatment success Definition: culture conversion on LJ medium at 8/52 136/176 (PP – 4 lost to follow-up) 44/58 10 mg/kg, 46/59 15 mg/kg, 46/59 20 mg/kg (PP) Treatment failure 5/178: 3/58 10 mg/kg, 1/59 15 mg/kg, 1/59 20 mg/kg (PP) Relapse 2/178: both at 10 mg/kg Death 0/180 Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2 115 Adverse events 80/180: 26/60 10 mg/kg, 31/60 15 mg/kg, 23/60 20 mg/kg Secondary outcomes of interest SASEs 9 in 8 participants Discontinuation 29/180 Drug-specific adverse events: DILI, rifamycin hypersensitivity syndromes Hepatoxicity: 16/60 10 mg/kg, 14/60 15 mg/kg, 14/60 20 mg/kg No episodes of hypersensitivity Thrombocytopenia not clearly documented as all haematological adverse events combined Disease specific efficacy outcomes of interest, i.e., extrapulmonary TB Not available Notes Authors: Peru, UK, USA Funding: US National Institute of Allergy and Infectious Diseases grants Yunivita et al., 2016 (13) Methods Open-label, randomized, three-arm study of patients with tuberculous meningitis. Randomized to: – 13 mg/kg RIF IV – 17 mg/kg RIF orally – 20 mg/kg RIF orally with standard HZE for 14/7, then onto standard treatment. Follow-up: unclear, probably only 2/52 Lost to follow-up: 0 Participants Setting: Hasan Sidikin Hospital, Bandung, Indonesia Number of participants: 30: 11 17 mg/kg, 9 20 mg/kg, 10 13 mg/kg Inclusion criteria: – Age ≥ 17 years – clinically suspected, untreated definite, probable or possible tuberculous meningitis Exclusion criteria: – treated for TB for ≥ 3/7 before admission Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2116 – ALT > 5x upper limit of normal – pregnancy – hypersensitivity to RIF – no diagnostic lumbar puncture – evidence of bacterial or cryptococcal meningitis HIV status: all tested, 6/30 positive: 1/11 17 mg/kg, 1/9 20 mg/kg, 4/10 13 mg/kg Baseline drug susceptibility: unclear. Methods state MTB drug resistance testing, but no results given. Intervention At randomization Study outcomes Primary outcomes: – PK data with different doses of RIF – safety and tolerability of different doses of RIF Primary outcomes of interest Treatment success Not a reported outcome Treatment failure Not a reported outcome Relapse Not documented Death 1/30 Adverse events 18/30 7/11 17 mg/kg, 7/9 20 mg/kg, 4/10 13 mg/kg Secondary outcomes of interest SAEs Not clear Discontinuation 1/30: 1/9 at 20 mg/kg Drug-specific adverse events: DILI, rifamycin hypersensitivity syndromes Hepatoxicity: – grade 3 – 8/30: 3/11 17 mg/kg, 3/9 20 mg/kg, 2/10 13 mg/kg – grade 4 – 1/30: 1/9 20 mg/kg Hypersensitivity: 0 No data on thrombocytopenia Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2 117 Disease-specific efficacy outcomes of interest, i.e., extrapulmonary TB None reported Notes Authors: Indonesia and the Netherlands Funding: This work was supported by a BOPTN—UNPAD grant from the Ministry of Education and Culture, Indonesia Abbreviations: ALT, Alanine Aminotransferase; ART, Antiretroviral therapy; AST, Aspartate Aminotransferase; CSF, cerebrospinal fluid; DILI, drug induced liver injury ; DOT, directly observed treatment; E, ethambutol; EFV; H, isoniazid; IQR, interquartile range; ITT, intention to treat; LJ, Lowenstein-Jensen; MDR, multidrug-resistant; MGIT, mycobacterial growth indicator tube; PK, pharmacokinetics; PP, per protocol; RCT, randomized controlled trial; R, rifampicin; RIF, rifampicin; UK, United Kingdom; USA, United States of America; Z, pyrazinamide; ZN, Ziehl-Neelsen Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2118 References 1. Aarnoutse RE, Kibiki GS, Reither K, Semvua HH, Haraka F, Mtabho CM et al. Pharmacokinetics, tolerability, and bacteriological response of rifampin administered at 600, 900, and 1,200 milligrams daily in patients with pulmonary tuberculosis. Antimicrob Agents Chemother. 2017;61(11):e01054-17. 2. Atwine D, Baudin E, Gelé T, Muyindike W, Mworozi K, Kyohairwe R et al. Effect of high-dose rifampicin on efavirenz pharmacokinetics: drug-drug interaction randomized trial. J Antimicrob Chemother. 2020;75(5):1250–8. 3. Boeree MJ, Heinrich N, Aarnoutse R, Diacon AH, Dawson R, Rehal S et al. High-dose rifampicin, moxifloxacin, and SQ109 for treating tuberculosis: a multi-arm, multi-stage randomised controlled trial. Lancet Infect Dis. 2017;17(1):39–49. 4. Dian S, Yunivita V, Ganiem AR, Pramaesya T, Chaidir L, Wahyudi K et al. Double-blind, randomized, placebo-controlled phase II dose-finding study to evaluate high-dose rifampin for tuberculous meningitis. Antimicrob Agents Chemother. 2018;62(12):e01014-18. 5. Heemskerk AD, Bang ND, Mai NT, Chau TT, Phu NH, Loc PP et al. Intensified antituberculosis therapy in adults with tuberculous meningitis. N Engl J Med. 2016;374(2):12–34. 6. Jindani A, Borgulya G, de Patiño IW, Gonzales T, de Fernandes RA, Shrestha B et al. A randomised phase II trial to evaluate the toxicity of high-dose rifampicin to treat pulmonary tuberculosis. Int J Tuberc Lung Dis. 2016;20(6):832–8. 7. Long MW, Snider DE Jr, Farer LS. US Public Health Service Cooperative trial of three rifampin-isoniazid regimens in treatment of pulmonary tuberculosis. Am Rev Respir Dis. 1979;119(6):879–94. 8. Maug AKJ, Hossain MA, Gumusboga M, Decroo T, Mulders W, Braet S et al. First-line tuberculosis treatment with double-dose rifampicin is well tolerated. Int J Tuberc Lung Dis. 2020;24(5):499–505. 9. Merle C, Floyd S, Ndiaye A, Galperine T, Furco A, De Jong B et al. High-dose rifampicin tuberculosis treatment regimen to reduce 12-month mortality of TB/HIV co-infected patients: the RAFA trial results. In: 21st International AIDS Conference. 18-22 July 2016, Durban, South Africa Abstract WEAB0205LB available at https://www.abstract-archive.org/ 10. Ruslami R, Nijland HM, Alisjahbana B, Parwati I, van Crevel R, Aarnoutse RE. Pharmacokinetics and tolerability of a higher rifampin dose versus the standard dose in pulmonary tuberculosis patients. Antimicrob Agents Chemother. 2007;51(7):2546–51. 11. Ruslami R, Ganiem AR, Dian S, Apriani L, Achmad TH, van der Ven AJ et al. Intensified regimen containing rifampicin and moxifloxacin for tuberculous meningitis: an open-label, randomised controlled phase 2 trial. Lancet Infect Dis. 2013;13(1):27–35. 12. Velásquez GE, Brooks MB, Coit JM, Pertinez H, Vargas Vásquez D, Sánchez Garavito E et al. Efficacy and safety of high-dose rifampin in pulmonary tuberculosis. A randomized controlled trial. Am J Respir Crit Care Med. 2018;198(5):657–66. 13. Yunivita V, Dian S, Ganiem AR, Hayati E, Hanggono Achmad T, Purnama Dewi A et al. Pharmacokinetics and safety/tolerability of higher oral and intravenous doses of rifampicin in adult tuberculous meningitis patients. Int J Antimicrob Agents. 2016;48(4):415–21. Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2 119 Appendix 3. Studies that were excluded from the review Study (reference) Reason for exclusion Atwine et al., 2018 (1) Poster presentation of included study by Atwine et al. (2020) Ding et al., 2020 (2) Pharmacokinetics study for included study by Heemskerk et al. (2016) Kuzhko et al., 2016 (3) Comparison between intravenous and oral administration and not between doses Kuzhko et al., 2017 (4) Comparison between intravenous and oral administration and not between doses Milstein et al., 2016 (5) Protocol for included study by Velasquez et al. (2018) Nageswari et al., 2018 (6) Wrong comparator: standard dose of rifampicin with a lower dose Peloquin et al., 2017 (7) Pharmacokinetics study for included study by Velasquez et al. (2018) Te Brake et al., 2015 (8) Pharmacokinetics study for included study by Ruslami et al. (2013) Van Crevel et al., 2012 (9) Oral presentation of included study by Ruslami et al. (2013) Velasquez et al., 2019(10) Duplicate of included study by Velasquez et al. (2018) References 1. Atwine D, Baudin E, Gele T, Muyindike WR, Kenneth M, Kyohairwe R et al. Efavirenz pharmacokinetics with rifampin double dose in TB-HIV infected patients. Top Antivir Med. 2018;26(suppl 1):185–6s. 2. Ding J, Thuy Thuong Thuong N, Pham TV, Heemskerk D, Pouplin T, Tran CTH et al. Pharmacokinetics and pharmacodynamics of intensive antituberculosis treatment of tuberculous meningitis. Clin Pharmacol Ther. 2020;107(4):1023–33. 3. Kuzhko M, Hulchuk N, Tlustova T, Avramchuk O, Gumeniuk M. The effectiveness of pulmonary TB treatment in patients with organic liver diseases, depending on the way of administration of anti-TB drugs. Eur Respir J. 2016;48:20160903–20160907. 4. Kuzhko M, Gumeniak M, Butov D, Tlustova T, Denysov O, Sprynsian T. Features of intravenous anti TB therapy in patients with first diagnosed pulmonary TB in the intensive phase of treatment (abstract). Eur Respir J. 2017 (doi:10.1183/1393003.congress-2017.PA3496). 5. Milstein M, Lecca L, Peloquin C, Mitchison D, Seung K, Pagano M et al. Evaluation of high-dose rifampin in patients with new, smear-positive tuberculosis (HIRIF): study protocol for a randomized controlled trial. BMC Infect Dis. 2016;16(1):453. 6. Nageswari AD, Rajanandh MG, Uday M, Nasreen RJ, Pujitha RR, Prathiksha G. Effect of rifampin with bio-enhancer in the treatment of newly diagnosed sputum positive pulmonary tuberculosis patients: a double-center study. J Clin Tuberc Other Mycobact Dis. 2018;12:73–7. Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2120 7. Peloquin CA, Velasquez GE, Lecca L, Calderon RI, Coit J, Milstein M et al. Pharmacokinetic evidence from the HIRIF trial to support increased dose of rifampin for tuberculosis. Antimicrob Agents Chemother. 2017;61(8):e00038-17. 8. Te Brake L, Dian S, Ganiem AR, Ruesen C, Burger D, Donders R et al. Pharmacokinetic/pharmacodynamic analysis of an intensified regimen containing rifampicin and moxifloxacin for tuberculous meningitis. Int J Antimicrob Agents. 2015;45(5):496–503. 9. Van Crevel R, Ruslami R, Ganiem AR, Dian S, Apriani L, Chaidir L et al. Pharmacokinetics, safety and effectiveness of high-dose rifampicin and moxifloxacin for tuberculosis meningitis: a randomized clinical trial in Indonesia. Clin Microbiol Infect. 2012;3:33–4. 10. Velásquez GE, Brooks MB, Coit JM, Sánchez Garavito E, Calderón RI et al (2019) Reply to te Brake et al : conflicting findings on an intermediate dose of rifampicin for pulmonary tuberculosis Am J Respir Crit Care Med. 2019 ; 199(9):1167-1168 Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2 121 Appendix 4. Funnel and Galbraith plots Treatment success 10 mg/kg versus 15 mg/kg 10 mg/kg versus 20 mg/kg All-cause mortality 10 mg/kg versus 15 mg/kg 10 mg/kg versus 20 mg/kg Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2122 Serious adverse events 10 mg/kg versus 15 mg/kg 10 mg/kg versus 20 mg/kg Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2 123 Appendix 5. Data and analyses Analysis 1. Treatment success: intention-to- treat analysis of risk ratio for rifampicin at doses of 10 mg/kg versus 15 mg/kg Analysis 2. Treatment success: intention-to- treat analysis of risk ratio for rifampicin at doses of 10 mg/kg versus 20 mg/kg Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2124 Analysis 3. Treatment success: per-protocol analysis of risk ratio for rifampicin at doses of 10 mg/kg versus 15 mg/kg and 10 mg/kg versus 20 mg/kg Analysis 4. Treatment failure: intention-to-treat analysis of risk ratio for rifampicin at doses of 10 mg/ kg versus 15 mg/kg and 10 mg/kg versus 20 mg/kg Analysis 5. Relapse: intention-to-treat analysis of risk ratio for rifampicin at doses of 10 mg/kg versus 15 mg/kg and 10 mg/kg versus 20 mg/kg Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2 125 Analysis 6. All-cause mortality: intention-to-treat analysis of risk ratio for rifampicin at doses of 10 mg/kg versus 15 mg/kg Analysis 7. All-cause mortality: intention-to-treat analysis of risk ratio for rifampicin at doses of 10 mg/kg versus 20 mg/kg Analysis 8. All-cause mortality: intention-to-treat analysis of risk ratio for rifampicin at doses of 10 mg/ kg versus 30 mg/kg and 10 mg/kg versus 35 mg/kg Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2126 Analysis 9. All-cause mortality: per-protocol analysis of risk ratio of rifampicin at doses of 10 mg/kg versus 15 mg/kg and 10 mg/kg versus 20 mg/kg Analysis 10. Serious adverse events: intention-to- treat analysis of risk ratio for rifampicin at doses of 10 mg/kg versus 15 mg/kg and 10 mg/kg versus 20 mg/kg Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2 127 Analysis 11. Drug-induced liver injury: intention-to- treat analysis of risk ratio for rifampicin at doses of 10 mg/kg versus 15 mg/kg Analysis 12. Drug-induced liver injury: intention-to- treat analysis of risk ratio for rifampicin at doses of 10 mg/kg versus 20 mg/kg Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2128 Analysis 13. Drug-induced liver injury: intention-to-treat analysis of risk ratio for rifampicin at doses of 10 mg/ kg versus 30 mg/kg and 10 mg/kg versus 35 mg/kg Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2 129 Appendix 6. Summary of findings Certainty assessment N° of patients Effect CertaintyN° of studies Study design Risk of bias Inconsist- ency Indirect- ness Impreci- sion Other considera- tions Higher doses of Rifampicin Standard doses of Rifampicin Relative (95% CI) Absolute (95% CI) Treatment Success 10 versus 15 mg/kg (ITT) 3 randomised trials not serious not serious not serious not serious none 392/459 (85.4%) 395/457 (86.4%) RR 1.00 (0.97 to 1.03) 0 fewer per 1,000 (from 26 fewer to 26 more) ⊕⊕⊕⊕ High Treatment Success 10 versus 20 mg/kg (ITT) 5 randomised trials not serious not serious not serious not serious none 552/651 (84.8%) 533/651 (81.9%) RR 1.01 (0.97 to 1.06) 8 more per 1,000 (from 25 fewer to 49 more) ⊕⊕⊕⊕ High Treatment Success 10 versus 35 mg/kg (ITT) 1 randomised trials not serious not serious not serious serious a none 59/63 (93.7%) 117/123 (95.1%) RR 0.98 (0.91 to 1.06) 19 fewer per 1,000 (from 86 fewer to 57 more) ⊕⊕⊕ Moderate Treatment failure 10 versus 15 mg/kg (ITT) 2 randomised trials not serious not serious not serious very seriousb none 7/309 (2.3%) 10/307 (3.3%) RR 0.71 (0.27 to 1.88) 9 fewer per 1,000 (from 24 fewer to 29 more) ⊕⊕ Low Treatment failure 10 versus 20 mg/kg (ITT) 2 randomised trials not serious not serious not serious very seriousb none 13/413 (3.1%) 11/408 (2.7%) RR 1.01 (0.29 to 3.61) 0 fewer per 1,000 (from 19 fewer to 70 more) ⊕⊕ Low Relapse 10 versus 15 mg/kg (ITT) 2 randomised trials not serious not serious not serious very seriousb none 5/310 (1.6%) 7/307 (2.3%) RR 0.79 (0.25 to 2.45) 5 fewer per 1,000 (from 17 fewer to 33 more) ⊕⊕ Low Relapse 10 versus 20 mg/kg (ITT) 2 randomised trials not serious not serious not serious very seriousb none 3/413 (0.7%) 3/408 (0.7%) RR 0.93 (0.07 to 12.72) 1 fewer per 1,000 (from 7 fewer to 86 more) ⊕⊕ Low Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2130 Certainty assessment N° of patients Effect CertaintyN° of studies Study design Risk of bias Inconsist- ency Indirect- ness Impreci- sion Other considera- tions Higher doses of Rifampicin Standard doses of Rifampicin Relative (95% CI) Absolute (95% CI) All-cause mortality 10 versus 15 mg/kg (ITT) 4 randomised trials not serious not serious not serious serious d none 22/459 (4.8%) 27/457 (5.9%) RR 0.80 (0.47 to 1.37) 12 fewer per 1,000 (from 31 fewer to 22 more) ⊕⊕⊕ Moderate All-cause mortality 10 to 20 mg/kg (ITT) 6 randomised trials not serious not serious not serious serious d none 17/671 (2.5%) 20/670 (3.0%) RR 0.93 (0.47 to 1.81) 2 fewer per 1,000 (from 16 fewer to 24 more) ⊕⊕⊕ Moderate Serious Adverse Events 10 versus 15 mg/kg (ITT) 3 randomised trials serious a not serious not serious very seriousf none 6/210 (2.9%) 3/207 (1.4%) RR 1.83 (0.50 to 6.69) 12 more per 1,000 (from 7 fewer to 82 more) ⊕ Very low Serious Adverse Events 10 versus 20 mg/kg (ITT) 6 randomised trials serious a not serious not serious serious g none 31/655 (4.7%) 35/653 (5.4%) RR 0.96 (0.53 to 1.76) 2 fewer per 1,000 (from 25 fewer to 41 more) ⊕⊕ Low Serious Adverse Events 10 versus 35 mg/kg (ITT) 1 randomised trials serious h not serious not serious very seriousa g none 4/59 (6.8%) 6/123 (4.9%) RR 1.30 (0.38 to 4.44) 15 more per 1,000 (from 30 fewer to 168 more) ⊕ Very low 4 randomised trials serious a seriousi not serious serious i none 19/277 (6.9%) 18/312 (5.8%) RR 0.99 (0.54 to 1.83) 1 fewer per 1,000 (from 27 fewer to 48 more) ⊕ Very low Drug-induced liver injury 10 versus 20 mg/kg (ITT) 7 randomised trials serious a not serious not serious serious i none 33/671 (4.9%) 33/670 (4.9%) RR 0.99 (0.63 to 1.56) 0 fewer per 1,000 (from 18 fewer to 28 more) ⊕⊕ Low Drug-induced liver injury 10 versus 30 mg/kg (ITT) 1 randomised trials serious h not serious not serious very seriousa k none 4/20 (20.0%) 3/20 (15.0%) RR 1.33 (0.34 to 5.21) 50 more per 1,000 (from 99 fewer to 632 more) ⊕ Very low Optimization of the dose of rifampicin in adults with presumed drug-susceptible tuberculosis: systematic review and meta-analysis Annex 2 131 Certainty assessment N° of patients Effect CertaintyN° of studies Study design Risk of bias Inconsist- ency Indirect- ness Impreci- sion Other considera- tions Higher doses of Rifampicin Standard doses of Rifampicin Relative (95% CI) Absolute (95% CI) Drug-induced liver injury 10 versus 35 mg/kg (ITT) 1 randomised trials serious h not serious not serious very seriousa k none 3/63 (4.8%) 1/123 (0.8%) RR 5.86 (0.62 to 55.17) 40 more per 1,000 (from 3 fewer to 440 more) ⊕ Very low CI: confidence interval; RR: risk ratio Explanations a. Only one study included in this analysis b. Treatment failure outcomes within strata well below Optimal Information Size and low numbers of events c. Relapse outcomes within strata well below Optimal Information Size and low numbers of events d. Mortality outcomes well below Optimal Information Size e. High risk of bias in 7 out of 14 included studies with regard to lack of blinding with regard to safety assessment f. SAE outcome well below Optimal Information Size and low numbers of events g. SAE outcome well below Optimal Information Size h. High risk of bias for safety outcomes as only one study included i. Direction of effect varies significantly between studies j. DILI outcome well below Optimal Information Size k. DILI outcome well below Optimal Information Size and low numbers of events Note: We computed the optimal information size for key outcomes from the pooled approximate event rates in the control arms and representative feasible effect sizes. Illustrative calculated optimal information sizes were as follows: Treatment success: 2216 for RR of 1.05 Treatment failure and relapse: 5034 for RR 1.5 Mortality, serious adverse events and drug-induced liver injury: 2941 for RR 1.5

Optimization of the dose of ethambutol in adults with presumed drug-susceptible tuberculosis: systematic review ANNEX 3 Firas Maghrabi1, Hannah Ryan2, Vittoria Lutje3, Sarah Nevitt2, Geraint Davies2 1 Royal Liverpool University Hospital, Liverpool, United Kingdom 2 University of Liverpool, Liverpool, United Kingdom 3 Liverpool School of Tropical Medicine, Liverpool, United Kingdom

Optimization of the dose of ethambutol in adults with presumed drug-susceptible tuberculosis: systematic review Annex 3 135 Contents 1. Background 137 1.1 Description of the condition 137 1.2 Description of the intervention 137 1.3 How the intervention might work 138 1.4 Why this review is important 138 1.5 Methods 139 2. Results 145 3. Conclusions 147 3.1 Summary of main results 147 3.2 Potential biases in the review process 148 3.3 Agreements and disagreements with other studies or reviews 148 4. Authors’ conclusions 149 4.1 Implications for practice 149 4.2 Implications for research 149 4.3 Acknowledgements 149 5. References 150 Appendix 1. Search strategy 152 Appendix 2. Excluded Studies 153 References 153

Optimization of the dose of ethambutol in adults with presumed drug-susceptible tuberculosis: systematic review Annex 3 137 1. Background 1.1 Description of the condition Tuberculosis (TB) remains the single biggest killer of all infectious diseases, with an estimated 10 million cases and 1.2 million deaths worldwide in 2018 (1). While 30 low- and middle-income countries account for almost 90% of this burden, a significant fraction can be attributed to risk factors such as HIV co-infection (9%) and diabetes (15%). While progress has recently been made in the control of TB, the annual decrease in incidence does not currently exceed 2%. Seven million people were treated with first-line TB regimens globally in 2018, with a treatment success rate of 85%, but estimates suggest that the case fatality rate among all incident cases was as high as 15%, with most of the deaths among people in the most economically active period of their lives (1). Although pulmonary disease is the commonest presentation of TB, extrapulmonary and disseminated disease, particularly with neurological involvement, may be associated with worse outcomes. In addition, some people with TB harbour resistant strains of Mycobacterium tuberculosis for which first-line therapy is not effective (2). Of those presenting with TB for the first time, 13.1% are resistant to isoniazid (INH) and 3.3% are resistant to rifampicin (RIF), requiring modified, longer, more expensive treatment regimens (3). 1.2 Description of the intervention First-line short-course regimens for TB have changed little since their introduction more than 40 years ago. The finding that when RIF and pyrazinamide (PZA) were added to INH the duration of therapy could be reduced to as little as 6 months established this trio of drugs as the backbone of effective therapy (4). A fourth drug was often used to prevent the emergence of resistance, particularly in people who already harboured INH-resistant strains. While streptomycin was initially widely used for this purpose, widespread resistance and the need for parenteral administration led to its replacement by the oral agent ethambutol (EMB), resulting in the current standard first- line regimen (5). EMB, discovered in 1961, is a synthetic inhibitor of arabinolyl transferase. Its mechanism of action involves disruption of incorporation of mycolic acids into the mycobacterial cell wall. EMB was first introduced for the treatment of TB in people with INH- and/or streptomycin-resistant TB or as part of re-treatment regimens. This experience led later to its use as an obvious substitute for streptomycin in the first-line regimen. Isolated resistance to EMB has not been associated with an increased risk of treatment failure during first-line treatment (6). EMB is also considered a useful component of regimens for multi-drug-resistant TB when susceptibility is demonstrated (7). The recommended dose of EMB is 15–25 mg/kg daily or 50 mg/kg for intermittent treatment regimens (although the latter has recently been deprecated (8)). EMB has high oral bioavailability and reliable absorption, and its primary route of elimination is renal (9). The pharmacokinetics of EMB vary substantially between subjects. It concentrates strongly (approximately 10 times) in the lesions of pulmonary TB (10), but concentrations in cerebrospinal fluid are typically less than half those in plasma and frequently below the wild-type minimum inhibitory concentration (11). The most important toxic effect of EMB is optic neuropathy, which occurs in 0.7–1.2% of patients at current doses and may result in permanent visual loss (12, 13); it is considered to be Optimization of the dose of ethambutol in adults with presumed drug-susceptible tuberculosis: systematic review Annex 3138 dose-related (14). EMB may also cause disturbance of liver enzyme activity, pruritis, arthralgia, gastrointestinal disturbance, headache and confusion. Few preclinical studies of concentration– or dose–response relations have been performed for EMB. In-vitro hollow-fibre studies suggest that the pharmacokinetics–pharmacodynamics index may be the area under the concentration–time curve (AUC) divided by the minimum inhibitory concentration (MIC) and that the maximum achievable rate of kill is 0.22 log10 cfu/mL per day (15). In early-phase clinical trials, the early bactericidal activity (EBA)0–2 of EMB at the currently recommended dose is 0.25 log10 cfu/mL per day and 0.177 log10 cfu/mL per day over the first 14 days of treatment (16), which compares favourably with the activity of many existing and novel anti-TB drugs. Selection of the currently recommended dose of EMB, however, appears to be based primarily on considerations of dose-limiting toxicity rather than efficacy. Modern first-line regimens rely on weight-based dosing strategies to control variations in pharmacokinetics and, after a prolonged period of pharmaceutical development, have been successfully co-formulated as fixed-dose combination tablets by a number of manufacturers (17). While truly individualized dosing approaches have been advocated, with therapeutic drug monitoring to identify individuals at the lower end of the distribution of pharmacokinetics, such strategies have not been widely implemented, even in settings with the required technical resources. For national TB programmes globally, altered dosing recommendations for all TB patients or for selected subgroups are the only ones that could be implemented immediately worldwide. For that reason, this review focuses on evidence suitable for universal or stratified rather than individualized dosing recommendations. 1.3 How the intervention might work The role of EMB in the first-line regimen has recently been re-evaluated. While it has been considered a drug with modest activity, used mainly to prevent the emergence of resistance, studies of pulmonary drug distribution suggest that unfavourable AUC/MIC values based on plasma measurements may not reflect its true contribution (10). The association between higher doses of the drug and optic neuropathy has, however, discouraged exploration of higher doses of EMB. The mechanism and determinants of this potentially serious toxic effect are not well understood, and it is not known whether ethambutol alternative dosing schedules might alter the risk–benefit of EMB in pulmonary or (particularly) meningeal TB. 1.4 Why this review is important Several lines of evidence suggest that strategies for dosing the individual drugs that comprise the current first-line regimen for TB may not be optimal. Maximizing the efficacy of the regimen could improve the long-term outcomes of TB treatment for all patients and make it more robust to variations in adherence, pharmacokinetics and pharmacogenetics and to the emergence of resistance. Furthermore, intensification of treatment with higher doses could be important for people with severe or disseminated disease and in populations particularly vulnerable to increased variation in pharmacokinetics or drug–drug interactions. Optimization of the dose of ethambutol in adults with presumed drug-susceptible tuberculosis: systematic review Annex 3 139 Objective To assess the efficacy and safety of doses higher than those currently recommended by WHO for each of the first-line anti-TB drugs (RIF, INH, PZA, EMB) used in a combination regimen for treating people with presumed drug-susceptible TB. 1.5 Methods The review was conducted according to the principles outlined in the PRISMA statement and PRISMA-P checklist (18, 19). Criteria for considering studies for this review Types of studies Randomized controlled trials Types of participants Inclusion criteria • Adults being treated for TB for the first time or being re-treated after a previous episode of TB Exclusion criteria • People being treated for TB with confirmed resistance to RIF, INH, PZA or EMB Adults are defined as people aged ≥ 18 years or treated as adults at participating centres in trials (inclusion of participants < 18 years to be clearly described). All forms of TB (pulmonary, extrapulmonary, disseminated) were considered for the purposes of the review, whether the diagnosis is based on bacteriological confirmation, imaging, biopsy or a clinical decision to treat. Type of interventions Intervention • Anti-TB treatment regimens containing EMB at doses higher than those recommended in current WHO guidelines (15–25 mg/kg daily) Comparator • Anti-TB treatment regimens containing EMB at doses recommended in current WHO guidelines The intervention was defined on the basis of comparisons based on dose, not on target concentrations in plasma, and randomized concentration-controlled trials were not included in the review. Although not recommended by WHO, administration of EMB by the intravenous route as an alternative to oral was permissible, provided that the target dose was different from those currently recommended. The dose metric used was actual weight-adjusted doses expressed in mg/kg. It was expected that many studies would not specify the target weight-adjusted dose in the report. When absolute doses were specified, the weight-adjusted dose for each trial was determined when possible from the description of weight data provided in the study report or, if these data were not Optimization of the dose of ethambutol in adults with presumed drug-susceptible tuberculosis: systematic review Annex 3140 available, the average weight in all the trials included in the analysis. When weight-banded dosing regimens were used, the average target weight-adjusted dose stated in the study report or the manufacturer’s summary of product characteristics was accepted, or, if not specified, the average of weight-adjusted doses computed from the mid-point of each band. Direct dose comparisons were defined as comparisons of different doses of EMB in the context of otherwise identical regimens (identical companion drugs, doses, duration and dosing intervals). Comparisons according to dose for each drug were considered between regimens of a similar composition and duration and regimens of any duration, whether greater than or less than 6 months, were considered potentially eligible. We did not consider direct comparisons between regimens of different doses and durations or between regimens with different doses and dosing intervals (daily versus less than daily). Types of outcome measures Primary outcomes We used definitions consistent with WHO programmatic outcomes (20) but adapted to permit evaluation of regimens longer or shorter than 6 months, of follow-up data in clinical trials and of trials of extrapulmonary TB. Treatment success In adults with pulmonary TB (bacteriologically confirmed, presumed drug-sensitive): smear- or culture-negative in the last month of treatment and on at least one previous occasion OR completed treatment without evidence of failure BUT with no record to show that sputum smear or culture results in the last month of treatment and on at least one previous occasion were negative, either because tests were not done or because results are unavailable. In adults with extrapulmonary TB: resolution of clinical signs and symptoms of TB at the end of treatment as judged by the investigators. Treatment failure In adults with pulmonary TB (bacteriologically confirmed, presumed drug-sensitive): sputum smear or culture positive within the last month of treatment. In adults with extrapulmonary TB: failure to resolve or return of clinical signs and symptoms of TB by the end of treatment as judged by the investigators Relapse In adults with pulmonary TB (bacteriologically confirmed, presumed drug-sensitive): smear or culture positive on one or more occasion during a defined period of follow-up after having previously met the criteria for treatment success. In adults with extrapulmonary TB: return of clinical signs and symptoms of TB as judged by the investigators after having previously met the criteria for treatment success Death Death from any cause before starting or during the course of treatment or during follow-up. Optimization of the dose of ethambutol in adults with presumed drug-susceptible tuberculosis: systematic review Annex 3 141 Adverse events All reported adverse events expressed as: total number of events and/or number of participants experiencing an event Secondary outcomes Serious adverse events All adverse events detailed in the study report that resulted in death, were life-threatening, required hospitalization or prolongation of hospitalization, resulted in persistent or significant disability or incapacity, consisted of a congenital anomaly or birth defect or were another important (protocol- defined) medical condition. Discontinuation Adverse events resulting in discontinuation of study medication Drug-specific adverse events of interest Optic neuropathy Disease-specific efficacy outcomes of interest Outcomes specific to trials in which people with extrapulmonary forms of TB were recruited, such as neurological disability in meningeal TB and pericardial constriction in pericardial TB. Search methods for identification of studies We identified all relevant studies, regardless of language, date of publication or publication status (published, unpublished, in press, in progress). Electronic searches We searched using the following electronic resources for eligible trials: MEDLINE, EMBASE, CENTRAL (Cochrane central register of controlled trials), Cochrane Infectious Diseases Group clinical trials register, WHO International Clinical Trials Registry and Clinicaltrials.gov We used the search strategy shown in Annex 1 (for CENTRAL, MEDLINE and EMBASE), which consists of the Cochrane Highly Sensitive Search Strategies for identification of clinical trials appropriate to those resources (21). Search for other resources We searched the reference lists of the retrieved study reports for any other studies. We also cross-referenced the results of the searches to the database of clinical trials of first-line TB treatment generated by our group for the systematic reviews conducted by the PreDiCT-TB consortium (22, 23). Data collection and analysis Selection of studies After de-duplication, the titles and abstracts of unique studies identified in the search were screened independently by two authors against the inclusion and exclusion criteria, and the full text was retrieved for all citations considered “eligible” or “unclear” but not for those considered “ineligible”. Two authors independently screened the retrieved full-text study reports, marked Optimization of the dose of ethambutol in adults with presumed drug-susceptible tuberculosis: systematic review Annex 3142 them as “eligible” or “ineligible” and recorded the reasons for exclusion in each case. Any disagreements were resolved by discussion with a third author. We used the Covidence web- based interface (24) to manage these processes and summarized the selection process in a PRISMA flow diagram. Data extraction and management We used a pilot-tested data extraction form for all the studies, which included the following items: Source: Lead author, year of publication, journal, PubMed ID, sponsor, trial registration number, corresponding author Methods: Study design, study dates and duration, sequence generation, allocation concealment, blinding, completeness of outcome data, selective reporting, other concerns about bias Participants: Number recruited, country, patient type (new/re-treatment), setting, inclusion/ exclusion criteria, age, sex, HIV status, diabetes status Microbiological methods: Details of diagnostic and susceptibility testing methods Pharmacology: Availability of data on pharmacokinetics and/or pharmacogenetics Interventions: Number of arms, names of drugs, doses of drugs, frequency of dosing, use of fixed-dose combinations, duration and structure of regimen, use of direct observation Outcomes: For each review outcome, we planned to extract the number randomized in each arm, the number included in the analysis and the number that experienced the event of interest to enable intention-to-treat, per-protocol and complete case analyses. We planned to record the duration of post-treatment follow-up in each case. Assessment of risk of bias Two review authors planned to independently assess the methodological quality of each included study with the Cochrane “Risk of bias” tool and to report the results in a “Risk of bias” table (25). Any disagreements would be resolved through discussion. Allocation sequence and allocation concealment were classified as adequate, inadequate or unclear in each included study according to Jüni (26). We planned to assess who was blinded in each included study and the risk of bias associated with blinding separately for each primary outcome. If at least 90% of participants were followed up to study completion, we would classify inclusion of all randomized participants as adequate; otherwise, we would classify inclusion as inadequate. We planned to attempt to contact the study authors for information that was unspecified or unclear. Measures of treatment effect We intended to use the risk ratio as the measure of treatment effect for analyses of both efficacy and safety. We also planned to express the results as the risk difference and number necessary to treat or harm, where appropriate. Missing data Analysis of the primary outcomes was planned to be on an intention-to-treat basis. If losses to follow-up in any of the studies were significant (> 10%), however, we planned to conduct a sensitivity analysis of the relevant trials to understand the impact of the missing data. Optimization of the dose of ethambutol in adults with presumed drug-susceptible tuberculosis: systematic review Annex 3 143 Assessment of heterogeneity We planned to assess heterogeneity by visual inspection of forest plots to determine the closeness of point estimates to each other and overlap of confidence intervals. For direct comparisons and meta-regression analyses, we planned to use the chi² test with a significance level of 0.10 (27) and the I² statistic (28) to assess heterogeneity, with a value of 50% taken to indicate significant statistical heterogeneity. For indirect comparisons, the τ2 statistic was also planned to be used to assess heterogeneity. Assessment of reporting biases We planned to conduct a visual inspection of funnel and Galbraith plots of the studies for any evidence of publication bias when at least 10 studies were included in the analysis. Data synthesis The review prioritized direct within-trial comparisons between doses of EMB that were part of otherwise similar regimens with the same total duration of treatment, to the extent permitted by the data. We aimed to perform meta-regression analyses of dose as a covariate to characterize dose–response relations for EMB as comprehensively as possible. We also aimed to graphically explore the possible additional information provided by indirect comparisons of doses between trials and perform a supplementary network meta-analysis, if supported by the structure of the dataset. Direct comparisons Meta-analysis, when appropriate, was planned to be performed with random effects models by the DerSimonian-Laird methods (29) or, if there were a few studies with low heterogeneity, fixed-effects models by the generalized inverse variance methods (27). When we wished to include more than one intervention study arm from a multi-armed study, we planned to split the control group to avoid including the same participants more than once. Stratified and unstratified forest plots would be used to present the data graphically. Corresponding 95% confidence intervals and P values would be computed, with a significance level of 0.05. We planned to quantify heterogeneity with the I2 statistic. To estimate overall dose–response relations from all the available data, meta-regression would be performed, with categorized and continuous weight- adjusted dose as a covariate and possibly other important sources of heterogeneity, such as duration and intermittent administration. Indirect comparisons For each drug, we aimed to categorize the treatment regimens used in the included studies in terms of their similarity (drugs included, dose levels, dosing schedule, structure and duration) and graphically explore the resulting network of treatments. If the structure of the network suggested an important contribution of indirect comparisons, we aimed to perform a network meta-analysis to incorporate this additional evidence. In addition to the updated effect estimates, we planned to compute the surface under the cumulative ranking curve and use the global I2 statistic and node-splitting to assess heterogeneity and inconsistency of the network. Statistical analyses would be conducted with OpenBUGS 3.2.3 (30) and R 3.4 (31) packages metafor, netmeta and gemtc. Optimization of the dose of ethambutol in adults with presumed drug-susceptible tuberculosis: systematic review Annex 3144 Subgroup analysis and investigation of heterogeneity The primary and secondary analyses specified above were planned to be stratified by the following subgroups as data permitted, for meaningful characterization and analysis: • People receiving treatment for the first time versus those previously treated • Pulmonary versus specific forms of extrapulmonary TB (for example meningeal TB) • Severity of disease (as measured by smear grade, radiological extent of disease, presence of cavitation, disseminated disease and/or mycobacteraemia) • HIV co-infection • Diabetes • Daily versus less-than-daily dosing schedule of the regimen (two or three times per week) • Weight-banded versus non-weight banded dosing strategies • Companion drugs in the regimen • Pharmacogenetic polymorphisms (as measured by phenotype or genotype) Certainty of the evidence We aimed to assess the certainty of the evidence with the GRADE approach (32). We planned to use GRADEpro GDT (33) to construct “Summary of findings” tables presenting ratings on certainty of evidence for effect estimates for each outcome, with relative and absolute measures of effect Sensitivity analysis We planned to perform the following sensitivity analyses when appropriate: (1) intention to treat, in which all participants who were randomized are included), (2) per protocol (in which only those participants who were randomized and completed the treatment course as planned are included), (3) worst-case scenario (in which participants who missed an outcome are assumed to have failed or relapsed for efficacy outcomes or had a negative safety outcome) and (4) best-case scenario (in which participants who missed an outcome are assumed to have had favourable efficacy and safety outcomes). Optimization of the dose of ethambutol in adults with presumed drug-susceptible tuberculosis: systematic review Annex 3 145 2. Results Of the 2457 citations screened, 2350 were excluded as they did not meet the inclusion criteria. Although 107 full-text articles were assessed for eligibility, 41 studies were excluded as of the wrong intervention, 60 were excluded as they had used the wrong comparator, 3 studies had the wrong study design, and 2 studies had the wrong study population; a further single study was identified as a duplicate. The reasons for exclusion for all of the retrieved studies are provided in the PRISMA diagram in Fig. 1. Fig. 1. PRISMA diagram of screened, excluded and included studies and reasons for exclusion Search N=4245 Title/Abstract Screening N=2457 Included in Qualitative Synthesis N=0 Included in Quantitative Synthesis N=0 Full-text Retrieval N=107 Eligible ETH N=5 Eligible RIF, INH, PZA or ETH N=38 Duplicates N=1788 Excluded N=2350 Excluded N=69 33 wrong intervention 30 wrong comparator 3 wrong study design 2 wrong population 1 duplicate Excluded N=5 4 wrong intervention 1 duplicate Optimization of the dose of ethambutol in adults with presumed drug-susceptible tuberculosis: systematic review Annex 3146 Five studies were initially identified as possible direct dose comparisons involving EMB, but one proved to be a duplicate publication. After further full-text screening, the four other studies were also excluded. One compared intravenous with oral administration of unspecified doses of EMB in people with pulmonary tuberculosis. Another compared intravenous administration of EMB, INH and “sodium rifamycin” with the same doses in mg/kg given orally to people with pulmonary TB. Butov et al. (34) compared intravenous with oral administration of isoniazid and EMB for the treatment of tuberculous meningitis in HIV-positive patients. The absolute dose in mg/kg of EMB was similar in both arms (although the bioavailability was different, resulting in different exposures, which was confirmed by pharmacokinetics analysis). The Tuberculosis Research Centre (35) compared four different EMB-containing regimens for treating pulmonary TB; although the absolute doses differed, a different dosing interval was used in each regimen, and the weekly cumulative dose in mg/kg was similar in all four groups. None of the identified studies was suitable for inclusion or further analysis (for details see Appendix 2 – Excluded Studies). Optimization of the dose of ethambutol in adults with presumed drug-susceptible tuberculosis: systematic review Annex 3 147 3. Conclusions 3.1 Summary of main results In this systematic review, none of the studies was relevant for addressing the question of whether higher doses of EMB than those currently recommended could safely improve outcomes in first- line treatment of adults with TB. Despite a sensitive, inclusive search strategy and extensive screening of abstracts and full texts, no randomized trials met the inclusion criteria. We therefore do not currently know whether higher doses of EMB would be of value in the treatment of TB. There may be several reasons for the lack of evidence. First, the review necessarily focused on modifications of the first-line regimen currently recommended by WHO, which is administered daily for 6 months by the oral route, usually as a fixed-dose combination. EMB has been used for the first 2 months of treatment in order to protect the bacterium from isoniazid resistance and has not been considered central to the overall efficacy of the regimen. Therefore, its independent contribution to treatment success does not seem to have been a priority for clinical trialists during the past 40 years, and clinical trials that included EMB before the 1980s do not conform closely to the characteristics of the currently accepted regimen. Secondly, while we did find variation in the absolute weight-based dose size of EMB in a number of studies (from 20 to 90 mg/kg), this was typically in the context of intermittent therapy with dosing schedules deliberately designed to administer an identical cumulative mg/kg dose over a week. The intent of these trials was clearly not to improve the efficacy of regimens but rather their operational feasibility, which was not the primary focus of this review. Thirdly, at current doses, EMB is known to be associated with an approximately 1% incidence of optic neuropathy, a serious, potentially irreversible side-effect, which may be dose-related. This may explain the reluctance of investigators to explore higher doses in clinical trials, as the likely risks may be perceived to outweigh the benefits, even in the context of serious forms of TB such as TB meningitis. EMB is known to distribute incompletely into cerebrospinal fluid, but the only trial that explored boosting of penetration of the meninges in TB meningitis (32) involved intravenous administration rather than increasing the absolute size of the dose. No clinical or pharmacokinetics data were provided to inform the apparent assumption of these investigators that a cumulative weekly dose of approximately 140 mg/kg is the acceptable limit of toxicity for the drug nor that cumulative dose is the relevant metric that determines the toxicity. Given these considerations, we do not anticipate that future clinical trials will directly address the question of EMB dosing, particularly as the role of the drug may change with the advent of rapid diagnostics for detection of INH resistance and the arrival of newer drugs that may allow shortening of the first-line regimen. While there is continuing debate about the role of EMB in TB meningitis, it appears likely that it will be replaced by other drugs, such as fluoroquinolones, in this context rather than be further evaluated in trials to optimize its dose. Observational studies, including of pharmacokinetics and toxicity, may increase understanding of the determinants of EMB toxicity and suggest alternative dosing strategies; however, they are unlikely to provide a basis for further trials, given the considerations stated above. Optimization of the dose of ethambutol in adults with presumed drug-susceptible tuberculosis: systematic review Annex 3148 3.2 Potential biases in the review process Our review has some limitations. It is possible that our search strategy failed to identify relevant studies; however, we included the broadest possible terms and screened a large number of potentially relevant studies. A more flexible analysis of a cumulative weekly dosing metric might have resulted in the inclusion of more studies and a broader range of doses; however, in many of the studies that we screened, the intermittent dosing strategies appeared to have been calibrated to a very similar cumulative weekly dose, resulting in relatively limited variation even in this metric. In addition, this approach would probably result in confounding between dose size and interval in the context of uncertainty about the pharmacokinetics–pharmacodynamics index of EMB. While we planned to explore the possibility of a network meta-analysis to comparing dose sizes by indirect comparisons, these objections would also apply to this approach and would be compounded in terms of heterogeneity by the variety of background regimens. 3.3 Agreements and disagreements with other studies or reviews We found no previous systematic reviews that specifically addressed dosing of EMB in adults. Optimization of the dose of ethambutol in adults with presumed drug-susceptible tuberculosis: systematic review Annex 3 149 4. Authors’ conclusions 4.1 Implications for practice We found no evidence to support the use of higher doses of EMB in the current first-line regimen for treatment of TB than those currently recommended by WHO. We also do not know whether subgroups of people with serious forms of TB, such as meningeal TB, or those with HIV co-infection might benefit from this intervention. 4.2 Implications for research New studies of the key pharmacological determinants of optic neuropathy due to EMB and clarification of its contribution to the efficacy of an evolving first-line regimen might inform the role and future use of the drug. 4.3 Acknowledgements We acknowledge the WHO Global TB Programme, which commissioned and funded this review and provided valuable guidance and comments during its preparation. Optimization of the dose of ethambutol in adults with presumed drug-susceptible tuberculosis: systematic review Annex 3150 5. References 1. 2019 Global tuberculosis report (WHO/CDS/TB/2019.15). Geneva: World Health Organization; 2019 (https://www.who.int/tb/publications/global_report/en/). 2. Espinal MA, Kim SJ, Suarez PG, Kam KM, Khomenko AG, Migliori GB et al. Standard short- course chemotherapy for drug-resistant tuberculosis: Treatment outcomes in 6 countries JAMA. 2000;283:2537–45 (doi: 10.1001/jama.283.19.2537). 3. 2020 Global tuberculosis report. Geneva: World Health Organization; 2020 (https://www.who.int/ publications/i/item/9789240013131). 4. Fox W, Ellard GA, Mitchison DA. Studies on the treatment of tuberculosis undertaken by the British Medical Research Council tuberculosis units, 1946–1986, with relevant subsequent publications. Int J Tuberc Lung Dis. 1999;3(10 Suppl 2):S231–79. 5. British Thoracic and Tuberculosis Association. Short-course chemotherapy in pulmonary tuberculosis. A controlled trial by the British Thoracic and Tuberculosis Association. Lancet. 1976;2(7995):1102–4. 6. Espinal MA, Kim SJ, Suarez PG, Kam KM, Khomenko AG, Migliori GB, Baéz J, Kochi A, Dye C, Raviglione MC. Standard short-course chemotherapy for drug-resistant tuberculosis: treatment outcomes in 6 countries. JAMA. 2000; 283(19):2537-45. 7. Consolidated guidelines for treatment of drug resistant tuberculosis (WHO/CDS/TB/2019.7). Geneva: World Health Organization; 2019 (https://www.who.int/tb/publications/2019/ consolidated-guidelines-drug-resistant-TB-treatment/en/). 8. Guidelines for treatment of drug-susceptible tuberculosis and patient care (2017 update) (WHO/HTM/ TB/2017.05). Geneva: World Health Organization; 2017 (https://www.who.int/tb/publications/2017/ dstb_guidance_2017/en/). 9. Peets EA, Sweeney WM, Place VA, Buyske DA. The absorption, excretion and metabolic fate of Ethambutol in man. Am Rev Respir Dis. 1965 91:51–8. 10. Zimmerman M, Lestner J, Prideaux B, O’Brien P, Dias-Freedman I, Chen C et al. Ethambutol parti- tioning in tuberculous pulmonary lesions explains its clinical efficacy. Antimicrob Agents Chemother. 2017;61:e00924-17. 11. Donald PR. Cerebrospinal fluid concentrations of antituberculosis agents in adults and children. Tuberc Edinb Scotl. 2010;90(5):279–92. 12. Chen SC, Lin MC, Sheu SJ. Incidence and prognostic factor of Ethambutol-related optic neuropathy: 10-year experience in southern Taiwan. Kaohsiung J Med Sci. 2015;31(7):358–62. 13. Yang HK, Park MJ, Lee JH, Lee CT, Park JS, Hwang JM. Incidence of toxic optic neuropathy with low-dose Ethambutol. Int J Tuberc Lung Dis. 2016;20(2):261–4. 14. Leibold JE. The ocular toxicity of Ethambutol and its relation to dose. Ann N Y Acad Sci. 1966;135(2):904–9. 15. Srivastava S, Musuka S, Sherman C, Meek C, Leff R, Gumbo T. Efflux-pump-derived multiple drug resistance to Ethambutol monotherapy in Mycobacterium tuberculosis and the pharmacokinetics and pharmacodynamics of Ethambutol. J Infect. 2010;201:1225–31. 16. Jindani A, Aber V, Edwards E, Mitchison D. The early bactericidal activity of drugs in patients with pulmonary tuberculosis. Am Rev Respir Dis. 1980;121:939–49. 17. Lienhardt C, Cook SV, Burgos M, Yorke-Edwards V, Rigouts L, Anyo G et al. Efficacy and safety of a 4-drug fixed-dose combination regimen compared with separate drugs for treatment of pulmonary tuberculosis: The Study C randomized controlled trial. JAMA. 2011;305:1415–23 (doi: 10.1001/ jama.2011.436). 18. Moher D, Liberati A, Tetzlaff J, Altman DG, The PRISMA Group Preferred Reporting Items for Systematic Reviews and Meta-Analyses. The PRISMA Statement. PLoS Med. 2009;6(7):e1000097 (https://doi. org/10.1371/journal.pmed.1000097). Optimization of the dose of ethambutol in adults with presumed drug-susceptible tuberculosis: systematic review Annex 3 151 19. Moher D, Shamseer L, Clarke M, Ghersi D, Liberati A, Petticrew M et al. Preferred Reporting Items for Systematic Review and Meta-Analysis Protocols (PRISMA-P) 2015 statement. Syst Rev. 2015;4:1 (doi: 10.1186/2046-4053-4-1). 20. Definitions and reporting framework for tuberculosis – 2013 revision (WHO/HTM/TB/2013.2). Geneva: World Health Organization; 2013 (https://apps.who.int/iris/bitstream/handle/10665/79199/9789241505345_ eng.pdf). updated December 2014 and January 2020 (WHO/HTM/TB/2013.2) Geneva, Switzerland: World Health Organization; 2017 21. Lefebvre C, Glanville J, Briscoe S, Littlewood A, Marshall C, Metzendorf MI et al. Technical supplement to Chapter 4: Searching for and selecting studies. In: Higgins JPT, Thomas J, Chandler J, Cumpston MS, Li T, Page MJ et al., editors. Cochrane handbook for systematic reviews of interventions, Version 6; 2019. 22. Bonnett LJ, Ken-Dror G, Koh GCKW, Davies GR. Comparing the efficacy of drug regimens for pulmonary tuberculosis: Meta-analysis of endpoints in early-phase clinical trials. Clin Infect Dis. 2017;65:46–54 (doi: 10.1093/cid/cix247). 23. Ken-Dror G, Bonnet LJ, Koh G, Davies GR. Comparing the efficacy of first-line drug regimens for pulmonary tuberculosis: meta-analysis of Phase III Trials (in preparation 2022) 24. Covidence systematic review software 2020. Melbourne: Veritas Health Innovation (www.covidence. org). 25. Higgins JP, Altman DG, Gøtzsche PC, Jüni P, Moher D, Oxman AD et al. The Cochrane Collaboration’s tool for assessing risk of bias in randomised trials. BMJ 2011;343:d5928 (doi: 10.1136/bmj.d5928). 26. Jüni P, Altman DG, Egger M. Systematic reviews in health care: Assessing the quality of controlled clinical trials. BMJ. 2001;323(7303):42–6. 27. Deeks JJ, Higgins JP, Altman DG, editors. Chapter 9: Analysing data and undertaking meta-analyses. In: Higgins JPT, Churchill R, Chandler J, Cumpston MS, editors. Cochrane handbook for systematic reviews of interventions, version 5.2.0 (updated June 2017). Cochrane; 2019 (www.training.cochrane. org/handbook). 28. Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. BMJ. 2003;327:557–60. 29. DerSimonian R, Laird N. Meta-analysis in clinical trials. Controlled Clin Trials. 1986;7:177–88. 30. OpenBUGS version 3.2.3. MRC Biostatistics Unit; 2014 (http://http://www.openbugs.net/w/ Downloads). 31. R Core Team. R: A language and environment for statistical computing. Vienna: R Foundation for Statistical Computing; 2018 (https://www.R-project.org/). 32. Guyatt G, Oxman AD, Vist G, Kunz R, Falck-Ytter Y, Alonso-Coelle P et al. GRADE: An emerging consensus on rating quality of evidence and strength of recommendations. BMJ. 2008;336:3. 33. GRADEpro GDT: GRADEpro Guideline Development Tool. Hamilton (ON): McMaster University; 2020 (developed by Evidence Prime, Inc.) (gradepro.org). 34. Butov D, Feshchenko Y, Kuzhko M, Gumenuik M, Yurko K, Grygorova A et al. Effectiveness of intravenous isoniazid and Ethambutol administration in patients with tuberculosis meningoencephalitis and HIV infection. Tuberc Resp Dis. 2019;82:1–8 (doi: 10.4046/trd.2019.0021). 35. Tuberculosis Research Centre, Indian Council of Medical Research. Ethambutol plus isoniazid for the treatment of pulmonary tuberculosis a controlled trial of four regimens. Tubercle. 1981;62(1):13–29. Optimization of the dose of ethambutol in adults with presumed drug-susceptible tuberculosis: systematic review Annex 3152 Appendix 1. Search strategy Search Set Central Medline Embase 1 Tuberculosis [tak] Tuberculosis [ms] Tuberculosis [mt] 2 Rifampicin [tak] OR Isoniazid [tak] OR Pyrazinamide [tak] OR Ethambutol [tak] Tuberculosis [tiab] Tuberculosis [tiab] 3 1 AND 2 1 OR 2 1 OR 2 4 Latent [tak] OR Prevention [tak] Rifampicin [tiab] OR Rifampin [tiab] OR Isoniazid [tiab] OR Pyrazinamide [tiab] OR Ethambutol [tiab] Rifampicin [tiab] OR Rifampin [tiab] OR Isoniazid [tiab] OR Pyrazinamide [tiab] OR Ethambutol [tiab] 5 3 NOT 4 Rifampin [sc] OR Isoniazid [sc] OR Pyrazinamide [sc] OR Ethambutol [sc] Rifampin [mt] OR Isoniazid [mt] OR Pyrazinamide [mt] OR Ethambutol [mt] 6 4 OR 5 4 OR 5 7 3 AND 6 3 AND 6 8 Randomized controlled trial [pt] OR Controlled clinical trial [pt] Crossover procedure [de] OR Double-blind procedure [de] OR Randomized controlled trial [de] OR Single-blind procedure [de] 9 Clinical trials as topic [mesh: no exp] Random* [deabti] OR Factorial* [deabti] OR Crossover* [deabti] OR Cross NEXT/1 over* [deabti] OR Placebo* [deabti] OR Doubl* NEAR/1 blind* [deabti] OR Singl* NEAR/1 blind* [deabti] OR Assign* [deabti] OR Allocat* [deabti] OR Volunteer* [deabti] 10 Randomized [tiab] OR Randomly [tiab] OR placebo [tiab] OR trial [ti] 8 OR 9 11 8 OR 9 OR 10 Animal experiment [mt] NOT (Human experiment [mt] OR Human [mt]) 12 Animals [mh] NOT Humans [mh] 10 NOT 11 13 11 NOT 12 7 AND 12 Optimization of the dose of ethambutol in adults with presumed drug-susceptible tuberculosis: systematic review Annex 3 153 Search Set Central Medline Embase 14 7 AND 13 Latent [tiab] OR Prevention [tiab] 15 Latent [tiab] OR Prevention [tiab] 13 NOT 14 16 14 NOT 15 [tak] word in title abstract or keyword, [pt] Publication Type, [ti ] word in title, [tiab] word in title or abstract, [sh] subheading, [mh] exploded MeSH term, [mesh: noexp] unexploded MeSH term, [mt] exploded Emtree term, [de] design, [deabti] Appendix 2. Excluded Studies Study (Reference) Reason for exclusion Butov 2019 (1) Wrong intervention Kushko 2016 (2) Wrong intervention Kushko 2017 (3) Wrong intervention and population Reusch 1978 (4) Wrong intervention TRC 1981 (5) Wrong intervention References 1. Butov D, Feshchenko Y, Kuzhko M, Gumenuik M, Yurko K, Grygorova A et al. Effectiveness of intravenous isoniazid and Ethambutol administration in patients with tuberculosis meningoencephalitis and HIV infection. Tuberc Resp Dis. 2019;82:1–8 (doi: 10.4046/trd.2019.0021). 2. Kuzhko M, Hulchuk N, Tlustova T, Avramchuk O, Gumeniuk M. The effectiveness of pulmonary TB treatment in patients with organic liver diseases, depending on the way of administration of anti-TB drugs. Eur Respir J. 2016;48:20160903–20160907. 3. Kuzhko M, Gumeniak M, Butov D, Tlustova T, Denysov O, Sprynsian T. Features of intravenous anti TB therapy in patients with first diagnosed pulmonary TB in the intensive phase of treatment (abstract). Eur Respir J. 2017 (doi:10.1183/1393003.congress-2017.PA3496). 4. Reusch G. Oral ethambutol treatment versus parenteral infusion therapy. Praxis und Klinik der Pneumologie 1978;32(12):791-801 5. Tuberculosis Research Centre, Indian Council of Medical Research. Ethambutol plus isoniazid for the treatment of pulmonary tuberculosis a controlled trial of four regimens. Tubercle. 1981;62(1):13–29.

Optimization of the dose of pyrazinamide in adults with presumed drug-susceptible tuberculosis: systematic review ANNEX 4 James Millard1, Sharon Isralls2, Hannah Ryan1, Sarah Nevitt1, Geraint Davies1 1 University of Liverpool, Liverpool, United Kingdom 2 Imperial NHS Foundation Trust, United Kingdom

Optimization of the dose of pyrazinamide in adults with presumed drug-susceptible tuberculosis: systematic review Annex 4 157 Contents 1. Background 158 1.1 Description of the condition 158 1.2 Description of the intervention 158 1.3 How the intervention might work 159 1.4 Why this review is important 159 1.5 Methods 159 2. Results 166 2.1 Description of studies 166 2.2 Studies excluded 168 3. Discussion 172 3.1 Summary of main results 172 3.2 Overall completeness and applicability of evidence 172 3.3 Certainty of the evidence 173 3.4 Potential biases in the review process 173 3.5 Agreements and disagreements with other studies or reviews 173 3.6 Authors’ conclusions 173 3.7 Acknowledgements 174 3.8 References 174 Appendix 1. Search strategy 177 Appendix 2. Studies that were excluded 179 References 179 Appendix 3. Summary of findings 180 Optimization of the dose of pyrazinamide in adults with presumed drug-susceptible tuberculosis: systematic review Annex 4158 1. Background 1.1 Description of the condition Tuberculosis (TB) remains the single biggest killer of all infectious diseases, with an estimated 10 million cases and 1.2 million deaths worldwide in 2018 (1). While 30 low- and middle-income countries account for almost 90% of this burden, a significant fraction can be attributed to risk factors such as HIV co-infection (9%) and diabetes (15%). While progress has recently been made in the control of TB, the annual decrease in incidence does not currently exceed 2%. Over seven million people were treated with first-line TB regimens globally in 2019 (1), with a treatment success rate of 85% in 2018, but estimates suggest that the case fatality rate among all incident cases was as high as 15%, with most of the deaths among people in the most economically active period of their lives (2). Although pulmonary disease is the commonest presentation of TB, extrapulmonary and disseminated disease, particularly with neurological involvement, may be associated with worse outcomes. In addition, some people with TB harbour resistant strains of Mycobacterium tuberculosis for which first-line therapy is not effective (3). Of those presenting with TB for the first time, 13.1% are resistant to isoniazid (INH) and 3.3% are resistant to rifampicin (RIF), requiring modified, longer, more expensive treatment regimens (1). 1.2 Description of the intervention First-line short-course regimens for TB have changed little since their introduction more than 40 years ago. The finding that when RIF and pyrazinamide (PZA) were added to INH the duration of therapy could be reduced to as little as 6 months established this trio of drugs as the backbone of effective therapy (4). A fourth drug was often used to prevent the emergence of resistance, particularly in people who already harboured INH-resistant strains. While streptomycin was initially widely used for this purpose, widespread resistance and need for parenteral administration led to its replacement by the oral agent ethambutol (ETH), resulting in the current standard first- line regimen (5). PZA is a synthetic analogue of nicotinamide and was first introduced for the treatment of TB in 1952. It was not established as an element of the first-line TB regimen until it was used in conjunction with RIF in pivotal clinical trials of short-course chemotherapy in the 1970s (4). In this context, when added to RIF, PZA was associated with an increase in rates of stable cure after a duration of treatment of 6 months (6). Trials of the duration of PZA in the regimen suggested that its contribution was largely confined to the first 2 months of therapy, with little additional efficacy afterwards (7, 8). Although rarely used or studied in the absence of RIF since that time, PZA is now also widely used in the treatment of multi-drug-resistant TB (9). The currently recommended dose of PZA is 25 (20-30) mg/kg daily, adjusted for weight according to banding schedules that depend on formulation as a single drug or as part of a fixed-dose combination (10). Higher daily doses were used in older clinical trials, and doses as high as 50-70 mg/kg have been used in intermittent treatment regimens (11). The rationale for choosing these doses was based on considerations of risk and benefit and particularly on the perceived risk of drug-induced liver injury due to PZA, although it is unclear whether this is dose-related (12). The tolerability of PZA may be limited by hyperuricaemia, itching, arthralgia and, less commonly, gout. Optimization of the dose of pyrazinamide in adults with presumed drug-susceptible tuberculosis: systematic review Annex 4 159 1.3 How the intervention might work Studies of the pharmacokinetics of PZA show that absorption is very reliable and plasma concentrations abundant (13). Although PZA penetrates modestly into tuberculous lesions, its metabolite pyrazinoic acid appears to accumulate two to three times inside them (14). The concentrations of PZA in cerebrospinal fluid are similar to those in plasma (15). The pharmacokinetics–pharmacodynamics target has not been established with certainty in preclinical or clinical studies but is suggested to be the area under the concentration–time curve and the maximum concentration (16). When used in the presence of RIF in short-course regimens, PZA reduced the risk of relapse, even when used only in the intensive phase (8). Whether this is due to simple pharmacological synergy between the two drugs or to targeting of PZA activity to subpopulations of M. tuberculosis not killed by RIF, for physiological or pathological reasons, remains unclear (17). Preclinical dose optimization of PZA has met with numerous technical problems. The drug is a weak base, and its activity may depend strongly on conditions of low pH. Hence, in vitro systems at neutral pH and animal models that do not reproduce human pathology may not reliably capture the activity of PZA or its concentration–response profile. In addition, in early-phase clinical trials, the early bactericidal activity of PZA is minimal, possibly due to its unusual mechanism of action (18). For this reason, unlike for RIF and INH, no early- phase dose titration studies have been performed, and there remains significant uncertainty about whether higher doses would improve efficacy, which would have to be established with definitive end-points. 1.4 Why this review is important The lines of evidence summarized above suggest that strategies for dosing PZA as part of the current first-line regimen for treatment for TB may not be optimal. Ensuring that maximized efficacy of the regimen could be important in improving the long-term outcomes of TB treatment for all patients and making it more robust to variations in adherence, pharmacokinetics and pharmacogenetics and to the emergence of resistance. Furthermore, intensification of treatment with higher doses could be important for people with severe or disseminated disease or in populations who are particularly vulnerable to increased variation in pharmacokinetics or drug– drug interactions. In addition to improved outcomes for individual TB sufferers, higher doses of PZA could ultimately reduce the duration of first-line treatment to less than 6 months, which would reduce the health system burden of treatment supervision and support and possibly better prevent the emergence of PZA resistance. Objective To assess the efficacy and safety of doses of PZA higher than those currently recommended by WHO when used as part of a combination regimen for treating people with presumed drug- susceptible TB 1.5 Methods This review was conducted according to the principles outlined in the PRISMA statement and PRISMA-P checklist (19, 20). Optimization of the dose of pyrazinamide in adults with presumed drug-susceptible tuberculosis: systematic review Annex 4160 Criteria for considering studies for this review Types of studies Randomized controlled trials Types of participants Inclusion criteria • Adults being treated for TB for the first time or being re-treated after a previous episode of TB Exclusion criteria • People being treated for TB with confirmed resistance to RIF, INH, PZA or ETH Adults were defined as people aged ≥ 18 years or treated as adults at participating centres in trials (inclusion of participants ≤ 18 years of age to be clearly described). All forms of TB (pulmonary, extrapulmonary and disseminated) were considered, whether the diagnosis was based on bacteriological confirmation, imaging, biopsy or a clinical decision to treat. Types of interventions Intervention • Anti-TB treatment regimens containing PZA at doses higher than those recommended in current WHO guidelines (25 mg/kg) Comparator • Anti-TB treatment regimens containing PZA at doses recommended in current WHO guidelines The intervention was defined on the basis of comparisons based on dose, not on target concentrations in plasma, and randomized concentration-controlled trials were not in scope of the review. Although not recommended in WHO guidance, administration of PZA by the intravenous route as an alternative to oral was permissible, provided that the target dose was different from those currently recommended. For the purposes of this review, the dose metric used was actual weight-adjusted doses expressed in mg/kg. It was expected that many study reports would not specify the target weight- adjusted dose. When absolute doses were specified, the weight-adjusted dose for each trial was determined as far as possible from the description of weight data provided in the study report or, if these data were not available, the average weight in all trials included in the analysis. When weight-banded dosing regimens were used, the average target weight-adjusted dose stated in the study report or the manufacturer’s summary of product characteristics was accepted, or, if it was not specified, the average of weight-adjusted doses computed from the mid-point of each band. For daily versus intermittent regimens, we considered the weekly cumulative weight- adjusted dose as the metric for analyses of efficacy and safety unstratified by these variables, where appropriate. Comparisons according to dose for each drug were considered for regimens of similar composition and duration and regimens of any duration, > or < 6 months, were considered potentially eligible. Optimization of the dose of pyrazinamide in adults with presumed drug-susceptible tuberculosis: systematic review Annex 4 161 Direct comparisons between regimens of different doses and durations were not considered in the primary analysis. Types of outcome measures Primary outcomes We used definitions consistent with WHO programmatic outcomes (21), adapted to permit evaluation of regimens longer or shorter than 6 months, of follow-up data in clinical trials and of trials of extrapulmonary TB. Treatment success In adults with pulmonary TB (bacteriologically confirmed, presumed drug-sensitive): smear- or culture-negative in the last month of treatment and on at least one previous occasion OR completed treatment without evidence of failure BUT with no record to show that sputum smear or culture results in the last month of treatment and on at least one previous occasion were negative, either because tests were not done or because results are unavailable. In adults with extrapulmonary TB: resolution of clinical signs and symptoms of TB at the end of treatment as judged by the investigators. Treatment failure In adults with pulmonary TB (bacteriologically confirmed, presumed drug-sensitive): sputum smear or culture positive within the last month of treatment. In adults with extrapulmonary TB: failure to resolve or return of clinical signs and symptoms of TB by the end of treatment as judged by the investigators. Relapse In adults with pulmonary TB (bacteriologically confirmed, presumed drug-sensitive): smear or culture positive on one or more occasion during a defined period of follow-up after having previously met the criteria for treatment success. In adults with extrapulmonary TB: return of clinical signs and symptoms of TB as judged by the investigators after having previously met the criteria for treatment success. Death Death from any cause before starting, during the course of treatment or during follow-up Adverse events All reported adverse events expressed as the total number of events and/or th number of participants who experienced an event. Secondary outcomes Serious adverse events All adverse events detailed in the study report that resulted in death, were life-threatening, required hospitalization or prolongation of hospitalization, resulted in persistent or significant disability or incapacity, consisted of a congenital anomaly or birth defect or were another important (protocol- defined) medical condition. Optimization of the dose of pyrazinamide in adults with presumed drug-susceptible tuberculosis: systematic review Annex 4162 Discontinuation Adverse events resulting in discontinuation of study medication Drug-specific adverse events of interest Drug-induced liver injury Arthralgia Disease-specific efficacy outcomes of interest Outcomes specific to trials that recruited people with extrapulmonary forms of TB, such as neurological disability in meningeal TB and pericardial constriction in pericardial TB. Search methods for identification of studies We identified all relevant studies, regardless of language, date of publication or publication status (published, unpublished, in press, in progress). Electronic searches We searched for eligible trials in the following electronic resources: MEDLINE, EMBASE, CENTRAL (Cochrane central register of controlled trials), Cochrane Infectious Diseases Group clinical trials register, WHO International Clinical Trials Registry and Clinicaltrials.gov. We used the search strategy described in Annex 1 (for CENTRAL, MEDLINE and EMBASE), which consist of the Cochrane Highly Sensitive Search Strategies for identification of clinical trials appropriate to those resources (22). Search for other resources We searched the reference lists of the retrieved study reports for any other studies. We contacted active investigators in the field to provide information on unidentified, ongoing or planned trials. We cross-referenced the results of the searches to the database of clinical trials of first-line TB treatment generated by our group for the systematic reviews conducted by the PreDiCT-TB consortium (11,23). Data collection and analysis Selection of studies After de-duplication, the titles and abstracts of unique studies identified in the search were screened independently by two authors against the inclusion and exclusion criteria, and the full text was retrieved for all citations considered “eligible” or “unclear” but not for those considered “ineligible”. Two authors independently screened the retrieved full-text study reports, marked them as “eligible” or “ineligible” and recorded the reasons for exclusion in each case. Any disagreements were resolved by discussion with a third author. We used the Covidence web- based interface (25) to manage these processes. Data extraction and management We designed and pilot-tested a data extraction form on a small subset of studies and modified it according to the results. After the form format was finalized and programmed in Covidence, two authors independently extracted data and compared the results. Any errors or discrepancies Optimization of the dose of pyrazinamide in adults with presumed drug-susceptible tuberculosis: systematic review Annex 4 163 were resolved by discussion and comparison with the study reports. We extracted the following data from each included study, when available: Source: Lead author, year of publication, journal, PubMed ID, sponsor, trial registration number, corresponding author Methods: Study design, study dates and duration, sequence generation, allocation concealment, blinding, completeness of outcome data, selective reporting, other concerns about bias Participants: Number recruited, country, patient type (new/retreatment), setting, inclusion/ exclusion criteria, age, sex, HIV status, diabetes status Microbiological methods: Details of diagnostic and susceptibility testing methods Pharmacology: Availability of data on pharmacokinetics and/or pharmacogenetics Interventions: Number of arms, names of drugs, doses of drugs, frequency of dosing, use of fixed-dose combinations, duration and structure of regimen, use of direct observation Outcomes: For each review outcome, we extracted the number of patients randomized to each arm, the number included in the analysis and the number that experienced the event of interest to enable intention-to-treat (ITT), per-protocol (PP) and best- and worst-case analyses. We recorded the duration of post-treatment follow-up in each case. The extracted data were be imported into R for Windows, version 4.0.3, for further management and analysis. Assessment of risk of bias Two review authors independently assessed the methodological quality of each included study using the Cochrane “Risk of bias” tool and reported the results in a “Risk of bias” table (26). We resolved any disagreements through discussion. To generate allocation sequence and allocation concealment, we classified each as either adequate, inadequate or unclear for each study according to Jüni et al. (27). We reported who was blinded in each included study and assessed the risk of bias associated with blinding separately for each primary outcome. If at least 90% of participants were followed up to study completion, we classified inclusion of all randomized participants as adequate; otherwise, we classified inclusion as inadequate. We attempted to contact the study authors for information that was unspecified or unclear. Measures of treatment effect We used the risk ratio (RR) as the measure of treatment effect for analyses of both efficacy and safety. Missing data Analysis of the primary outcomes was on an ITT basis, with missing outcomes assumed to be adverse. If losses to follow-up in any of the studies were significant (> 10%), however, we planned to conduct a sensitivity analysis of the relevant trials to determine the impact of the missing data. Assessment of heterogeneity We assessed heterogeneity by visual inspection of forest plots to determine the closeness of point estimates to each other and overlap of confidence intervals (CIs). For direct comparisons and meta-regression analyses, we used the chi² test with a significance level of 0.10 (28) and the I² Optimization of the dose of pyrazinamide in adults with presumed drug-susceptible tuberculosis: systematic review Annex 4164 statistic (29) to assess heterogeneity, with a value of 50% taken to indicate significant statistical heterogeneity. For indirect comparisons, the τ2 statistic was also used to assess heterogeneity. Assessment of reporting biases We planned to conduct visual inspection of Funnel and Galbraith plots of the studies for any evidence of publication bias if at least ten studies were included in the analysis for an outcome. Data synthesis We prioritized direct within-trial comparisons between doses of PZA that were part of otherwise similar regimens for the same total duration of treatment. We planned to perform meta-regression analyses of dose as a covariate to characterize dose–response relations for each of the drugs as comprehensively as possible. We also planned to perform a supplementary network meta- analysis, incorporating indirect comparisons, if supported by the structure of the dataset. Direct comparisons Meta-analysis, where appropriate, was planned to be performed with random effects models with the DerSimonian-Laird method (30) in the first instance, as we anticipated heterogeneity among the included studies due to differences in background regimen composition, year of study, geographical location, disease severity and co-morbid conditions. If there were few studies with minimal differences in design and low heterogeneity, fixed-effects models with generalized inverse variance method (28) were planned. For multi-arm studies, when we wished to include more than one intervention study arm, we planned to divide the control group to avoid including the same participants more than once. Stratified and unstratified forest plots were planned to present the data graphically, with corresponding 95% confidence intervals (CIs) and P values, computed with a significance level of 0.05. We quantified heterogeneity with the I2 statistic. To estimate overall dose–response relations in all the available data, meta-regression performed with categorized and continuous weight-adjusted dose as a covariate was planned. Indirect comparisons For each of the drugs, we planned to categorize the treatment regimens used in the studies in terms of their similarity (drugs included, dose levels, dosing schedule, structure and duration) and to explore the resulting network of treatments graphically. If the structure of the network suggested an important contribution of indirect comparisons, we planned to perform a network meta-analysis to incorporate this additional evidence. In addition to the updated effect estimates, we planned to compute the surface under the cumulative ranking curve (SUCRA) and to use the global I2 statistic and node-splitting to assess heterogeneity and inconsistency of the network. Statistical analyses were conducted in R 3.4 (31) with the package metafor. Subgroup analysis and investigation of heterogeneity The primary and secondary analyses specified above were planned to be stratified by the following subgroups when the data permitted meaningful characterization and analysis: • People receiving treatment for the first time versus those previously treated • Pulmonary versus specific forms of extrapulmonary TB (for example meningeal TB) • Severity of disease (as measured by smear grade, radiological extent of disease, presence of cavitation, disseminated disease and/or mycobacteraemia) • HIV co-infection Optimization of the dose of pyrazinamide in adults with presumed drug-susceptible tuberculosis: systematic review Annex 4 165 • Diabetes • Daily versus less-than-daily dosing schedule of the regimen (two or three times per week) • Weight-banded versus non-weight banded dosing strategies • Companion drugs in the regimen • Pharmacogenetic polymorphisms (phenotypic or genotypic acetylator status) Certainty of the evidence We assessed the certainty of the evidence with the GRADE approach (32). We planned to use GRADEpro GDT (33) to construct “Summary of findings” tables presenting ratings on certainty of evidence for effect estimates for each outcome with relative and absolute measures of effect. Sensitivity analysis We planned to perform the following sensitivity analyses when appropriate: (i) ITT, in which all participants who were randomized are included); (ii) PP, in which only those participants who were randomized and completed the treatment course as planned are included); (iii) worst- case scenario, in which participants who missed an outcome are assumed to have failed or relapsed for efficacy outcomes or had a negative safety outcome; and (iv) best-case scenario, in which participants who missed an outcome are assumed to have had favourable efficacy and safety outcomes. Optimization of the dose of pyrazinamide in adults with presumed drug-susceptible tuberculosis: systematic review Annex 4166 2. Results 2.1 Description of studies Results of the search We screened 2457 citations. One publication of one trial (780 participants) was suitable for inclusion. The reasons for inclusion and exclusion are shown in the PRISMA diagram (Fig. 1). Studies included A single trial met the inclusion criteria. The trial was conducted in 1959 by the US Public Health Service TB therapy trial group (34). At the time of the study, the antimycobacterial activity of PZA had been confirmed in murine models (35–37) and patients (38–40), and the ability of PZA to sterilize infected tissues had been confirmed (37). Several of these clinical studies had, however, identified apparent hepatotoxicity during PZA administration (38, 39–41), albeit with apparently variable prevalence and severity (1.1% with mainly mild symptoms to 9.8%, including jaundice and fulminant hepatitis). The rationale of the trial included in this review was therefore to include a control group without PZA, different doses of PZA, different total durations of therapy and more patients than in previous trials with PZA. It should be noted that the regimens used in the trial were very different from those used for modern short-course treatment and did not include RIF. Geographical location and period The trial was conducted in 20 hospitals in the USA between 1954 and 1956. Fig. 1. PRISMA diagram Optimization of the dose of pyrazinamide in adults with presumed drug-susceptible tuberculosis: systematic review Annex 4 167 Search N=4245 Title/Abstract Screening N=2457 Included in Qualitative Synthesis N=1 Included in Quantitative Synthesis N=0 Full-text Retrieval N=107 Eligible PZA N=3 Eligible RIF, INH, PZA or ETH N=38 Duplicates N=1788 Excluded N=2350 Excluded N=69 33 wrong intervention 30 wrong comparator 3 wrong study design 2 wrong population 1 duplicate Excluded N=2 2 wrong intervention Optimization of the dose of pyrazinamide in adults with presumed drug-susceptible tuberculosis: systematic review Annex 4168 Participants Participants were included if they had microbiologically confirmed TB and excluded if they had “recognized liver disease”. The majority (84.2%) had already been treated for TB before inclusion in the trial; 91.1% were exposed to streptomycin, 88.4% to isoniazid and 92.2% to para-aminosalicylic acid (PAS). Only 21.3% of participants had first been admitted to care < 1 year previously, and 95% had radiological evidence of cavitation. These characteristics suggest an inpatient cohort with relatively advanced disease and substantial experience of treatment with inadequate treatment response. The study authors themselves describe the study population as “made up largely of ‘treatment failures’”. Interventions Participants were randomised to one of five regimens of 24 weeks total duration: • 24 weeks INH and PAS (n=152) • 24 weeks INH and 25 mg/kg PZA (n=194) • 12 weeks INH and 25 mg/kg PZA followed by 12 weeks INH and PAS (n=122) • 24 weeks INH and 40 mg/kg PZA (n=197) • 12 weeks INH and 40 mg/kg PZA followed by 12 weeks INH and PAS (n=115) INH and PAS were dosed at 4 mg/kg and 10 mg/kg, respectively, in all arms. Follow-up Participants were retained on trial for the full duration of treatment, but no post-treatment follow-up was reported. Outcome measures No efficacy outcomes of interest were reported. The safety outcomes reported included treatment discontinuation, drug-induced liver injury, mortality from all causes and mortality from TB. No data were reported on total adverse events, serious adverse events or arthralgia. A number of laboratory tests of liver function were performed, including 1-min and total serum bilirubin, bromsulfthalein and cephalin flocculation tests. 2.2 Studies excluded We excluded 109 studies at full text screening for the following principal reasons: non- randomized design (4), wrong intervention (31), no direct dose comparison (72) or non-identical background regimen. Risk of bias in the included study The trial was conducted well before reporting guidelines for randomized controlled trials were in place. Consequently, some of the information necessary to assess risk of bias was incompletely described. A summary of the risk of bias is provided in Fig. 2. Optimization of the dose of pyrazinamide in adults with presumed drug-susceptible tuberculosis: systematic review Annex 4 169 Fig. 2. Risk of bias in the study that was included Sequence generation and allocation concealment Randomization processes were not described in full, and bias could not be excluded. Randomization was apparently performed by the central study office, although the method was not specified. Local investigators were notified of participants’ initial treatment assignment by telegram, and the duration of PZA dosing was not known until 12 weeks of treatment were completed. Two centres agreed only to randomize participants to the 24-week INH and PZA arms, a restriction on randomization that led to numerical imbalance between the arms and possible selection bias. Blinding Neither participants nor personnel were blinded to treatment allocation; therefore, performance bias is possible. Participants in different arms could have been treated differently, for instance being more likely to be considered for some outcomes if in a particular arm. Three of the four reported outcomes were either objective unambiguous clinical outcomes, for example death, or blinded to those assessing the outcome, for example drug-induced liver injury. Blinding of outcome assessment was therefore deemed “adequate” overall. The discontinuation outcome may, however, have been liable to detection bias, as this could not be blinded to treatment arm. Incomplete outcome data Retention in the trial was > 90%, suggesting a low risk of attrition bias. Selective reporting Although no protocol was available for the trial, the trial report suggests that apparently pre- specified outcomes were reported by the investigators, reducing the risk of reporting bias. Other Publication bias could not be assessed on the basis of the single published trial available, but, given the contemporary clinical salience of the trial question, the possibility that there are other unpublished trials dating from this period cannot be excluded. From the event rates observed in the single trial and with conventional parameters of α=0.05 and β=0.20, even with a large effect size corresponding to a 50% reduction in relative risk, the optimal information size for the included outcomes ranged from 974 to 1272. As this was considerably larger than the actual sample size of the trial, we downgraded the evidence by one level for Imprecision. Optimization of the dose of pyrazinamide in adults with presumed drug-susceptible tuberculosis: systematic review Annex 4170 Effects of interventions As only one relevant trial was identified for this review, no formal data synthesis was performed. Instead, comparisons were made in the following two groupings of the trial arms: (i) 24 weeks of INH and PZA; 25 mg/kg PZA vs 40 mg/kg PZA and (ii) 12 weeks of INH and PZA, followed by 12 weeks of INH and PAS; 25 mg/kg PZA vs 40 mg/kg PZA. Data from the two 25-mg/kg and the two 40-mg/kg PZA arms were then combined to provide a relative measure of effect between the two dose levels. The results of comparing 25 mg/kg to 40 mg/kg PZA in otherwise identical arms and combining arms with the same dose of PZA, as described above, are presented in Table 1. The interpretation was based on what we would now describe as a PP analysis; however, we re-analysed the data on a modified ITT (mITT) and best-case basis as specified in the protocol. The three analyses are presented together. The PP and best-case estimates are similar, while the mITT and worst-case estimates are more conservative and closer to the null. In the following section, we discuss the PP estimates, which appear to be more appropriate for the safety-related outcomes. Table 1. Relative effect (risk ratio, RR) of 40 mg/kg vs 25 mg/kg PZA in a regimen of 6 months INH and PZA (6HZ), 3 months INH and PZA followed by 3 months INH and PAS (3HZ/3HP) or the two regimens combined, for various safety outcomes Outcome Per-protocol Best-case Modified intention to treat and worst-case 6 HZ 3 HZ/3HP Combined 6HZ 3HZ/3HP Combined 6HZ 3HZ/3HP Combined Discontinu- ation 4.79 [1.07– 21.53] 1.50 [0.49–4.58] 2.39 [0.79–7.28] 4.77 [1.06–21.47] 1.50 [0.49–4.52] 2.39 [0.79–7.24] 1.74 [0.86–3.51] 1.38 [0.63–3.0] 1.56 [0.93–2.64] Drug- induced liver injury 2.63 [0.86–8.10] 0.72 [0.12–4.19] 1.65 [0.49–5.62] 2.63 [0.85–8.09] 0.71 [0.12–4.14] 1.64 [0.48–5.63] 1.54 [0.80–2.98] 1.06 [0.41–2.72] 1.36 [0.79–2.34] All-cause mortality 1.09 [0.41–2.95] 1.29 [0.41–4.08] 1.17 [0.55–2.49] 1.09 [0.40–2.94] 1.27 [0.40–4.04] 1.16 [0.55–2.47] 1.07 [0.57–2.04] 1.27 [0.58–2.81] 1.15 [0.70–1.89] TB mortality 1.68 [0.50–5.62] 1.34 [0.37–4.85] 1.51 [0.63–3.64] 1.67 [0.50–5.60] 1.33 [0.37–4.79] 1.50 [0.62–3.62] 1.25 [0.63–2.49] 1.30 [0.56–2.99] 1.27 [0.74–2.16] When the trial arms were combined, the discontinuation rates at the higher dose of PZA appeared to be higher, but the evidence was very uncertain (RR 2.39 [0.7–7.28], RD 4% [1–8%], 1 trial, 628 participants). The trend to increased discontinuation appeared to be driven by an apparently large effect in the 6HZ arms (RR 4.79 [1.07–21.53], RD 5% [1–9%], 1 trial, 391 participants), with a smaller effect in the 3HZ/3HP arms (RR 1.50 [0.49–4.58], RD 3% [–4–10%], 1  trial, 237 participants). The presence of drug-induced liver injury was assessed from weekly tests for 1-min bilirubin, total bilirubin and cephalin flocculation testing; values above the upper limit of normal were reported for 28%, 6% and 24%, respectively. It is therefore not clear how these tests were used to identify clinically important drug-induced liver injury during treatment. The rates of drug-induced liver injury might have been higher with higher doses of PZA in the 6HZ arms only (RR 2.63 [0.86–8.10], RD 4% [0–9%], 1 trial, 391 participants), but they might have been be lower in the 3HZ/3HP arms (RR 0.72 [0.12–4.19], RD –1% [–5–4%], 1 trial, 237 participants), although in both cases the evidence was very uncertain. All-cause mortality (RR 1.17 [0.55– 2.49], RD 1% [–3–5%], 1 trial, 628 participants) and TB mortality (RR 1.51 [0.63–3.64], RD 2% Optimization of the dose of pyrazinamide in adults with presumed drug-susceptible tuberculosis: systematic review Annex 4 171 [–2–5%]) might have been increased with higher doses of PZA, but the evidence is very uncertain. For all outcomes, the absolute measures of effect of higher and lower doses of PZA were small, irrespective of background regimen (Table 2). Table 2. Absolute effect (risk difference; RD) of 40 mg/kg vs 25 mg/kg PZA in a regimen of 6 months INH and PZA (6HZ), 3 months INH and PZA followed by 3 months INH and PAS (3HZ/3HP) or the two regimens combined, for various safety outcomes Outcome Per-protocol Best-case mITT and worst-case 6HZ 3HZ/3HP Combined 6HZ 3HZ/3HP Combined 6HZ 3HZ/3HP Combined Discontinu- ation 0.05 [0.01–0.09] 0.03 [–0.04–0.1] 0.04 [0.01–0.08] 0.05 [0.01–0.09] 0.03 [–0.04–0.1] 0.04 [0.01–0.08] 0.05 [–0.01–0.11] 0.04 [–0.05–0.13] 0.04 [–0.01–0.1] Drug- induced liver injury 0.04 [0.0–0.09] –0.01 [–0.05–0.04] 0.02 [–0.03–0.07] 0.04 [0.0–0.08] –0.01 [–0.05–0.03] 0.02 [–0.03–0.06] 0.04 [–0.02–0.11] 0.0 [–0.07–0.08] 0.03 [–0.02–0.08] All-cause mortality 0.0 [–0.04–0.05] 0.01 [– 0.05–0.08] 0.01 [–0.03–0.05] 0.0 [–0.04–0.05] 0.01 [–0.05–0.08] 0.01 [–0.03–0.04] 0.01 [–0.05–0.07] 0.03 [–0.06–0.12] 0.01 [–0.04–0.07] TB mortality 0.02 [–0.02–0.06] 0.01 [–0.04–0.08] 0.02 [–0.02–0.05] 0.02 [–0.02–0.05] 0.01 [–0.04–0.07] 0.01 [–0.02–0.05] 0.02 [–0.04–0.08] 0.03 [–0.05–0.11] 0.02 [–0.03–0.07] Important data not found We found no trials that allowed us to explore efficacy outcomes with different doses of PZA in “otherwise identical” regimens. The single trial that reported safety outcomes was over 60 years old. We had planned to perform meta-regression analyses with dose as a covariate; however, we found only one set of dose comparisons (25 mg/kg vs 40 mg/kg), and this was not possible. Almost all the clinical trial data that support current use of PZA are from studies of regimens that include INH and, in most cases, INH and RIF (11). It is possible that the safety profile of different doses of PZA would be different in modern regimens. We found no trials that allowed us to explore this. We found no eligible trials that allowed assessment of the role of different doses of PZA within regimens for INH- or multi-drug resistant TB. None of the planned subgroup analyses was possible, including assessment of the effect of HIV infection on efficacy and safety outcomes with different doses of PZA. Although we have formed a network of trials to allow network meta- analysis, incorporating direct and indirect comparisons, once our inclusion criteria were applied, only the trial included here with direct comparisons was eligible. Optimization of the dose of pyrazinamide in adults with presumed drug-susceptible tuberculosis: systematic review Annex 4172 3. Discussion 3.1 Summary of main results We identified a single randomized controlled trial, published more than 60 years ago, that directly addressed the question of PZA dosing when part of the first-line treatment regimen for TB. We found no statistically significant differences in treatment discontinuation, drug-induced liver injury or death at a dose of 40 mg/kg of PZA from those with the currently favoured dose of 25 mg/kg. Although the relative risk of discontinuation or drug-induced liver injury was numerically higher in the arms containing PZA, the risk appeared to depend on the background regimen and duration of PZA dosing. The trial was open-label, and the methods fr assessing hepatotoxicity were not clearly described, suggesting an important risk of bias in these data. The possibly higher risk of discontinuation of treatment with 40 mg/kg rather than 25 mg/kg of PZA, when used in a regimen with INH in which both were given for 6 months but not in a regimen in which this combination was given for only 3 months and then PZA switched to PAS, might suggest that higher doses of PZA increase the risk over time when given for a longer period. In other trials with formal comparisons of different durations of PZA in diverse regimens, however, no effect was seen on the risk of adverse events, suggesting that much of the toxicity of PZA occurs early in treatment, with minimal additional risk during prolonged treatment (7, 8). Most of the randomized clinical trials that included PZA that were reviewed did not directly compare the doses of PZA between arms, often because higher doses of PZA were reserved for less-than-daily dosing, typically resulting in a similar weekly cumulative dose. A less common reason for exclusion was that arms were not “otherwise identical” in terms of companion drugs. This suggests that identification of the correct PZA dose has rarely been directly addressed in a clinical trial. This may be due to the fact that, in the 1950s, when PZA was developed, it was suggested to be associated with significant hepatoxicity at doses up to 50 mg/kg (38– 41) and was considered to be unsuitable for routine use for some time. In subsequent clinical trials in the 1970s, a lower dose of PZA was used, ultimately resulting in PZA being perceived as indispensable in current regimens (4). Those trials were not, however, supported by formal dose-ranging studies or use of modern preclinical techniques. Once treatment shortening by adding PZA to RIF-containing regimens was established, with acceptable toxicity at lower doses of 30–40 mg/kg, equipoise on the question was clearly disturbed and it is likely that assessment of the dose of PZA used in clinical trials was no longer seen as a priority. 3.2 Overall completeness and applicability of evidence The single trial included in the review was conducted more than 60 years ago and reported only on mortality and safety outcomes. The participants were highly treatment-experienced in-patients with TB, who may well represent target populations of people on first-line treatment. Only two dose levels of PZA (25 mg/kg and 40 mg/kg) were evaluated in two slightly different background regimens, neither of which contained RIF. As RIF is the foundation of modern first-line therapy and PZA has always been used in regimens including it since the 1980s, the evidence presented here must be considered highly incomplete and of limited applicability to guide changes in PZA dose in the current first-line regimen. Optimization of the dose of pyrazinamide in adults with presumed drug-susceptible tuberculosis: systematic review Annex 4 173 3.3 Certainty of the evidence The conclusions of this review are based on data from a single randomized controlled trial in a population that was clinically different from the target population of people receiving first-line treatment for TB. No efficacy outcomes were reported. We were concerned about the methods used in the trial in terms of randomization and blinding in relation to assessment of some outcomes. We therefore assessed the overall level of evidence as very low for the safety of higher doses of PZA and found no evidence to determine whether the intervention could improve the efficacy of the current first-line regimen (Annex 2). 3.4 Potential biases in the review process We conducted the review according to our agreed protocol and excluded studies for which direct comparisons of absolute dose of PZA were possibly confounded by intermittent administration or by background regimens containing different companion drugs or doses of those drugs in order to avoid bias. This narrow focus might have restricted our view of the evidence, and we were unable to conduct a feasible network meta-analysis, which would have enabled use of any information from indirect comparisons. The few poorly connected set of trials that we retrieved might simply reflect a lack of trials on the question of the optimal dose of PZA. Our search strategy was broad and inclusive, and it is unlikely that we missed any important published studies; however, we cannot rule out publication bias among older, smaller trials of PZA. 3.5 Agreements and disagreements with other studies or reviews Pasipanodya et al. (12) conducted a systematic review of the toxicity of PZA; however, their search strategy was insensitive and was not restricted to randomized trials, and a number of studies that the search strategy failed to identify were added to the dataset. The studies had diverse regimens, including PZA monotherapy for latent TB infection, and the meta-analysis presented showed very high levels of heterogeneity (I2 > 95%). The authors studied only safety outcomes. They found no difference in the rates of hepatotoxicity in regimens with and without PZA overall but noted increasing rates of hepatotoxicity with doses of 30, 40 and ≥ 60 mg/kg, which they did not find to be statistically significant. Their conclusions are broadly in agreement with those of this review, although our analysis is probably less vulnerable to bias and confounding. We did not identify any other systematic reviews on this question. 3.6 Authors’ conclusions Implications for practice Guidelines contemporaneous with the clinical trials that established modern first-line therapy recommended a daily dose of PZA of 35 mg/kg, which was typically used in those studies. The currently recommended daily dose of 25 mg/kg (range, 20–30 mg/kg) was introduced in the WHO guidelines in 2003 (42), but with no explicit rationale for the change. This weight-adjusted dose has subsequently been adopted in all available fixed-dose formulation products. This review found no evidence for greater efficacy of higher doses of PZA, and the data on safety from a single trial were very uncertain. Nevertheless, the rates of discontinuation, biochemically defined Optimization of the dose of pyrazinamide in adults with presumed drug-susceptible tuberculosis: systematic review Annex 4174 drug-induced liver injury and mortality might have been higher with higher doses of PZA, although this may also depend on the background regimen. Current evidence does not therefore support modification of the currently recommended dose of PZA in first-line TB treatment regimens. Implications for research This review focused narrowly on direct comparisons of dose in clinical trials, and we therefore did not use information from indirect comparisons to overcome the limitations in the available evidence. We suggest that the currently recommended dose be reviewed in a broader approach, incorporating indirect and pharmacokinetics– pharmacodynamics evidence when possible and new research when available. Work is under way to better define the exposure–response relation and potential treatment-shortening with higher exposures to PZA from observational data (43), but further phase-II clinical trials may be required to clearly define the maximally tolerated dose of PZA, as has been done for RIF (44, 45), before any larger, phase-III trials of direct dose comparisons of PZA are contemplated. 3.7 Acknowledgements We acknowledge the WHO Global TB Programme, which commissioned and funded this review and provided valuable guidance and comments during its preparation 3.8 References 1. Global tuberculosis report 2020. Geneva: World Health Organization; 2020 (https://www.who.int/ publications/i/item/9789240013131). 2. Global tuberculosis report 2019 (WHO/CDS/TB/2019.15). Geneva: World Health Organization; 2019 (https://apps.who.int/iris/bitstream/handle/10665/329368/9789241565714-eng.pdf). 3. Espinal MA, Kim SJ, Suarez PG, Kam KM, Khomenko AG, Migliori GB et al. Standard short- course chemotherapy for drug-resistant tuberculosis: treatment outcomes in 6 countries JAMA. 2000;283:2537–45 (doi: 10.1001/jama.283.19.2537). 4. Fox W, Ellard GA, Mitchison DA. Studies on the treatment of tuberculosis undertaken by the British Medical Research Council tuberculosis units, 1946–1986, with relevant subsequent publications. Int J Tuberc Lung Dis. 1999;3(10 Suppl 2):S231–79. 5. British Thoracic and Tuberculosis Association. Short-course chemotherapy in pulmonary tuberculosis. A controlled trial by the British Thoracic and Tuberculosis Association. Lancet. 1976;ii(7995):1102–4. 6. East African/British Medical Research Council. Controlled clinical trial of four short course regimens of chemotherapy for two durations in the treatment of pulmonary tuberculosis Am Rev Resp Dis. 1978;118:39–48. 7. Singapore Tuberculosis Service/British Medical Research Council. Clinical trial of 6-month and 4-month regimens of chemotherapy in the treatment of pulmonary tuberculosis. Am Rev Resp Dis. 1979;119:579–85. 8. Hong Kong Chest Service/British Medical Research Council. Controlled trial of 2, 4, and 6 months of pyrazinamide in 6 month, three-times-weekly regimens for smear positive pulmonary tuberculosis, including an assessment of a combined preparation of isoniazid, rifampin, and pvrazinamide Am Rev Resp Dis. 1991;143:700–6. 9. WHO Treatment guidelines for drug-resistant tuberculosis: 2016 update (WHO/HTM/TB/2016.04). Geneva: World Health Organization; 2016 (https://www.who.int/publications/i/item/9789241549639). Optimization of the dose of pyrazinamide in adults with presumed drug-susceptible tuberculosis: systematic review Annex 4 175 10. Guidelines for treatment of drug-susceptible tuberculosis and patient care (2017 update) (WHO/HTM/ TB/2017.05). Geneva: World Health Organization; 2017 (https://www.who.int/tb/publications/2017/ dstb_guidance_2017/en/). 11. Ken-Dror G, Bonnet LJ, Koh G, Davies GR. Comparing the efficacy of first-line drug regimens for pulmonary tuberculosis: meta-analysis of phase III trials (in preparation 2022) 12. Pasipanodya JG, Gumbo T. Clinical and toxicodynamic evidence that high-dose pyrazinamide is not more hepatotoxic than the low doses currently used. Antimicrob Agents Chemother. 2010;54:2847– 54 (doi: 10.1128/AAC.01567-09). 13. Chirehwa MT, McIlleron H, Rustomjee R, Mthiyane T, Onyebujoh P, Smith P et al. Pharmacokinetics of pyrazinamide and optimal dosing regimens for drug-sensitive and -resistant tuberculosis. Antimicrob Agents Chemother. 2017;61:e00490-17. 14. Via LE, Savic R, Weiner DM, Zimmerman MD, Prideaux B, Irwin SM et al. Host-mediated bioactivation of pyrazinamide: implications for efficacy, resistance and therapeutic alternatives. ACS Infect Dis. 2015;1(5):203–14. 15. Pouplin T, Bang ND, Toi PV, Phuong PN, Dung NH, Duong TN et al. Naïve-pooled pharmacokinetic analysis of pyrazinamide, isoniazid and rifampicin in plasma and cerebrospinal fluid of Vietnamese children with tuberculous meningitis. BMC Infect Dis. 2016;16:144. 16. Gumbo T, Siyambalapitiyage Dona C, Meek C, Leff R. Pharmacokinetics-pharmacodynamics of pyrazinamide in a novel in vitro model of tuberculosis for sterilizing effect: a paradigm for faster assessment of new antituberculosis drugs. Antimicrob Agents Chemother. 2009;53:3197–204. 17. Mitchison DA. Basic mechanisms of chemotherapy. Chest. 1979;76(S6):771–81. 18. Diacon AH, Dawson R, von Groote-Bidlingmaier F, Symons G, Venter A, Donald PR et al. Bactericidal activity of pyrazinamide and clofazimine alone and in combinations with pretomanid and bedaquiline. Am J Respir Crit Care Med. 2015;191:943–53 (doi: 10.1164/rccm.201410-1801OC). 19. Moher D, Liberati A, Tetzlaff J, Altman DG, the PRISMA Group. Preferred Reporting Items for Systematic Reviews and Meta-Analyses: The PRISMA statement. PLoS Med. 2009;6(7):e1000097 (https://doi. org/10.1371/journal.pmed.1000097). 20. Moher D, Shamseer L, Clarke M, Ghersi D, Liberati A, Petticrew M et al. Preferred Reporting Items for Systematic Review and Meta-Analysis Protocols (PRISMA-P) 2015 statement. Syst Rev. 2015;4:1 (doi: 10.1186/2046-4053-4-1). 21. Definitions and reporting framework for tuberculosis – 2013 revision (WHO/HTM/TB/2013.2). Geneva: World Health Organization; 2013 (https://apps.who.int/iris/bitstream/handle/10665/79199/9789241505345_ eng.pdf). 22. Lefebvre C, Glanville J, Briscoe S, Littlewood A, Marshall C, Metzendorf MI et al. Technical supplement to Chapter 4: Searching for and selecting studies. In: Higgins JPT, Thomas J, Chandler J, Cumpston MS, Li T, Page MJ et al., editors. Cochrane handbook for systematic reviews of interventions, version 6. Oxford: Cochrane; 2019. 23. Bonnett LJ, Ken-Dror G, Koh GCKW, Davies GR. Comparing the efficacy of drug regimens for pulmonary tuberculosis: meta-analysis of endpoints in early-phase clinical trials. Clin Infect Dis. 2017;65:46–54 (doi: 10.1093/cid/cix247). 24. Ken-Dror G, Davies GR, Bonnet LJ, Nevitt SJ. Exploring optimal first-line treatment for tuberculosis: a network meta-analysis (in preparation 2022) 25. Covidence systematic review software 2020. Melbourne: Veritas Health Innovation; 2020 (www. covidence.org). 26. Higgins JP, Altman DG, Gøtzsche PC, Jüni P, Moher D, Oxman AD et al. The Cochrane Collaboration’s tool for assessing risk of bias in randomised trials. BMJ 2011;343:d5928 (doi: 10.1136/bmj.d5928). 27. Jüni P, Altman DG, Egger M. Systematic reviews in health care: assessing the quality of controlled clinical trials. BMJ 2001;323(7303):42–6. Optimization of the dose of pyrazinamide in adults with presumed drug-susceptible tuberculosis: systematic review Annex 4176 28. Deeks JJ, Higgins JP, Altman DG, editors, on behalf of the Cochrane Statistical Methods Group. Chapter 9: Analysing data and undertaking meta-analyses. In: Higgins JPT, Churchill R, Chandler J, Cumpston MS, editors, Cochrane handbook for systematic reviews of interventions, version 5.2.0 (updated June 2017). Oxford: Cochrane; 2019 (www.training.cochrane.org/handbook). 29. Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. BMJ. 2003;327:557–60. 30. DerSimonian R, Laird N. Meta-analysis in clinical trials. Controlled Clin Trials. 1986;7:177–88. 31. R Core Team. R: A language and environment for statistical computing. Vienna: R Foundation for Statistical Computing; 2018 (https://www.R-project.org/). 32. Guyatt G, Oxman AD, Vist G, Kunz R, Falck-Ytter Y, Alonso-Coelle P et al. GRADE: an emerging consensus on rating quality of evidence and strength of recommendations. BMJ. 2008;336:3. 33. GRADEpro GDT: GRADEpro Guideline Development Tool. Hamilton (ON): McMaster University; 2020 (gradepro.org). 34. United States Public Health Service. Tuberculosis therapy trial. Hepatic toxicity of pyrazinamide used with isoniazid in tuberculous patients. Am Rev Resp Dis. 1959;80(3):371–87. 35. Dessau F, Yeager R, Burger F, Williams J. Pyrazinamide (aldinamide) in experimental tuberculosis of the guinea pig. Am Rev Tuberc. 1952;65(5):519–22. 36. Malone L, Schurr A, Lindh D, McKenzie D, Kiser J, Williams J. The effect of pyrazinamide (aldinamide) on experimental tuberculosis in mice. Am Rev Tuberc. 1952;65(5):511–8. 37. Mccune R, McDermott W, Tompsett R. The fate of Mycobacterium tuberculosis in mouse tissues as determined by the microbial enumeration technique. II. The conversion of tuberculous infection to the latent state by the administration of pyrazinamide and a companion drug. J Exp Med. 1956;104(5):763– 802 (doi:10.1084/jem.104.5.763). 38. Schwartz W, Moyer R. The chemotherapy of pulmonary tuberculosis with pyrazinamide used alone and in combination with streptomycin, para-aminosalicylic acid, or isoniazid. Am Rev Tuberc. 1954;70(3):413–22 (doi: 10.1164/art.1954.70.3.413). 39. Yeager R, Munroe W, Dessau F. Pyrazinamide (Aldinamide) in the treatment of pulmonary tuberculosis. Am Rev Tuberc. 1952;65(5):523–46. 40. McDermott W, Ormond L, Muschenheim C, Deuschle K, McCune R, Tompsett R. Pyrazinamide– isoniazid in tuberculosis. Am Rev Tuberc. 1954;69(3):319–33 (doi: 10.1164/art.1954.69.3.319). 41. Campagna M, Calix A, Hauser G. Observations on the combined use of pyrazinamide (Aldinamide) and isoniazid in the treatment of pulmonary tuberculosis; a clinical study. Am Rev Tuberc. 1954;69(3):334– 50 (doi: 10.1164/art.1954.69.3.334). 42. Treatment of tuberculosis: guidelines for national programmes (WHO/CDS/TB/2003.313). Geneva: World Health Organization; 2003 (http://apps.who.int/iris/bitstream/handle/10665/67890/WHO_CDS_ TB_2003.313_eng.pdf). 43. Svensson E, Svensson J, Te Brake L, Boeree M, Heinrich N, Konsten S et al. The potential for treatment shortening with higher rifampicin doses: relating drug exposure to treatment response in patients with pulmonary tuberculosis. Clin Infect Dis. 2018;67(1):34–41 (doi: 10.1093/cid/ciy026). 44. Boeree MJ, Diacon AH, Dawson R, Narunsky K, du Bois J, Venter A et al. A dose-ranging trial to optimize the dose of rifampin in the treatment of tuberculosis. Am J Respir Crit Care Med. 2015;191:1058–65. 45. Jindani A, Borgulya G, de Patiño IW, Gonzales T, de Fernandes RA, Shrestha B et al. A randomised phase II trial to evaluate the toxicity of high-dose rifampicin to treat pulmonary tuberculosis. Int J Tuberc Lung Dis. 2016;20:832–8 Optimization of the dose of pyrazinamide in adults with presumed drug-susceptible tuberculosis: systematic review Annex 4 177 Appendix 1. Search strategy Search Set Central Medline Embase 1 Tuberculosis [tak] Tuberculosis [ms] Tuberculosis [mt] 2 Rifampicin [tak] OR Isoniazid [tak] OR Pyrazinamide [tak] OR Ethambutol [tak] Tuberculosis [tiab] Tuberculosis [tiab] 3 1 AND 2 1 OR 2 1 OR 2 4 Latent [tak] OR Prevention [tak] Rifampicin [tiab] OR Rifampin [tiab] OR Isoniazid [tiab] OR Pyrazinamide [tiab] OR Ethambutol [tiab] Rifampicin [tiab] OR Rifampin [tiab] OR Isoniazid [tiab] OR Pyrazinamide [tiab] OR Ethambutol [tiab] 5 3 NOT 4 Rifampin [sc] OR Isoniazid [sc] OR Pyrazinamide [sc] OR Ethambutol [sc] Rifampin [mt] OR Isoniazid [mt] OR Pyrazinamide [mt] OR Ethambutol [mt] 6 4 OR 5 4 OR 5 7 3 AND 6 3 AND 6 8 Randomized controlled trial [pt] OR Controlled clinical trial [pt] Crossover procedure [de] OR Double-blind procedure [de] OR Randomized controlled trial [de] OR Single-blind procedure [de] 9 Clinical trials as topic [mesh: no exp] Random* [deabti] OR Factorial* [deabti] OR Crossover* [deabti] OR Cross NEXT/1 over* [deabti] OR Placebo* [deabti] OR Doubl* NEAR/1 blind* [deabti] OR Singl* NEAR/1 blind* [deabti] OR Assign* [deabti] OR Allocat* [deabti] OR Volunteer* [deabti] 10 Randomized [tiab] OR Randomly [tiab] OR placebo [tiab] OR trial [ti] 8 OR 9 11 8 OR 9 OR 10 Animal experiment [mt] NOT (Human experiment [mt] OR Human [mt]) 12 Animals [mh] NOT Humans [mh] 10 NOT 11 13 11 NOT 12 7 AND 12 Optimization of the dose of pyrazinamide in adults with presumed drug-susceptible tuberculosis: systematic review Annex 4178 Search Set Central Medline Embase 14 7 AND 13 Latent [tiab] OR Prevention [tiab] 15 Latent [tiab] OR Prevention [tiab] 13 NOT 14 16 14 NOT 15 [tak] word in title abstract or keyword, [pt] Publication Type, [ti ] word in title, [tiab] word in title or abstract, [sh] subheading, [mh] exploded MeSH term, [mesh: noexp] unexploded MeSH term, [mt] exploded Emtree term, [de] design, [deabti] Optimization of the dose of pyrazinamide in adults with presumed drug-susceptible tuberculosis: systematic review Annex 4 179 Appendix 2. Studies that were excluded Study Reason for exclusion African MRC 1969 (1) Wrong intervention African MRC 1970 (2) Wrong intervention References 1. East African/British Medical Research Council. A controlled comparison of four regimens of streptomycin plus pyrazinamide in the retreatment of pulmonary tuberculosis. Tubercle (London) 1969 ; 50 : 81-114 2. East African/British Medical Research Council. Streptomycin plus pyrazinamide in the retreatment of pulmonary tuberculosis -second report. Tubercle (London) 1970 ; 51 : 359-374 Optimization of the dose of pyrazinamide in adults with presumed drug-susceptible tuberculosis: systematic review Annex 4180 Appendix 3. Summary of findings Certainty assessment N° of patients Effect CertaintyN° of studies Study design Risk of bias Inconsist- ency Indirect- ness Impreci- sion Other considera- tions Higher doses of PZA Standard doses of PZA Relative (95% CI) Absolute (95% CI) Treatment discontinuation (ITT) 1 randomised trials serious a not serious serious b seriousc none 33/277 (11.9%) 21/285 (7.4%) RR 1.56 (0.93 to 2.64) 41 more per 1,000 (from 5 fewer to 121 more) ⊕ Very low Drug-induced liver injury (ITT) 1 randomised trials serious a not serious serious b seriousc none 29/277 (10.5%) 21/285 (7.4%) RR 1.36 (0.79 to 2.34) 27 more per 1,000 (from 15 fewer to 99 more) ⊕ Very low All-cause mortality (ITT) 1 randomised trials serious a not serious serious b seriousc none 30/277 (10.8%) 26/285 (9.1%) RR 1.15 (0.70 to 1.69) 14 more per 1,000 (from 27 fewer to 63 more) ⊕ Very low TB Mortality (ITT) 1 randomised trials serious a not serious serious b seriousc none 28/277 (10.1%) 22/285 (7.7%) RR 1.27 (0.74 to 2.16) 21 more per 1,000 (from 20 fewer to 90 more) ⊕ Very low Treatment Discontinuation (PP) 1 randomised trials serious a not serious serious b seriousc none 17/261 (6.5%) 7/271 (2.6%) RR 2.39 (0.79 to 7.28) 36 more per 1,000 (from 5 fewer to 162 more) ⊕ Very low Drug-induced liver injury (PP) 1 randomised trials serious a not serious serious b seriousc none 13/261 (5.0%) 7/271 (2.6%) RR 1.65 (0.49 to 5.62) 17 more per 1,000 (from 13 fewer to 119 more) ⊕ Very low All-cause mortality (PP) 1 randomised trials serious a not serious serious b seriousc none 14/261 (5.4%) 12/271 (4.4%) RR 1.17 (0.55 to 2.49) 8 more per 1,000 (from 20 fewer to 66 more) ⊕ Very low Optimization of the dose of pyrazinamide in adults with presumed drug-susceptible tuberculosis: systematic review Annex 4 181 Certainty assessment N° of patients Effect CertaintyN° of studies Study design Risk of bias Inconsist- ency Indirect- ness Impreci- sion Other considera- tions Higher doses of PZA Standard doses of PZA Relative (95% CI) Absolute (95% CI) TB Mortality (PP) 1 randomised trials serious a not serious serious b seriousc none 12/261 (4.6%) 8/271 (3.0%) RR 1.51 (0.63 to 3.54) 15 more per 1,000 (from 11 fewer to 75 more) ⊕ Very low CI: confidence interval; RR: risk ratio Explanations a. Randomisation and concealment methods were unclear and trial was open label b. The study regimen did not certain RIF. The study population consisted of treatment experienced patients, he majority of whom likely remained drug-susceptible though this was not confirmed by testing within the study. Their disease and physical status may not be representative of the target population of treatment-naive people with TB. c. Single trial with 628 participants (well under Optimal Information Size)

For further information, please contact: World Health Organization 20, Avenue Appia CH-1211 Geneva 27, Switzerland Global TB Programme Web site: www.who.int/tb