Subgroup considerations Subgroups included in the recommendation The panel suggested that the shorter regimen can be used in the subgroups for which evidence was available for review (people living with HIV infection, persons with diabetes mellitus, those with a low body weight and those with extensive disease). However, the panel also emphasized that additional research on the use of the shorter regimen in these subgroups is desirable. People living with HIV infection: The proportion of patients living with HIV infection in the intervention and control regimen arms was 8% and only patients with CD4 count above 100 cells/mm3 were enrolled. Of all the persons with HIV who participated in the trial (in all three arms), 95.4% were receiving antiretroviral treatment (ART). HIV-positive persons not on ART at enrollment, had planned initiation of efavirenz-based ART before or at study week 8. Persons with HIV were excluded from enrollment in the trial if, at the time of enrollment, their CD4-T cell count was known to be <100 cells/mm3 . Overall, there were nine patients who were not on ART throughout the trial follow-up in the microbiologically-eligible analysis population (4.6%); the reasons for non-initiation of ART were not clear. People with diabetes mellitus: Additional information from pharmacokinetic analyses will be available for this population in the future which may provide more nuanced evidence on the use of the intervention and control regimens in persons with diabetes mellitus. People with extensive TB disease: The trial reported on the presence of cavitation on chest radiograph (CXR),the extent of disease on CXR as a percentage, and cavity size (absent, < or > = 4cm). For patients with less severe and minimal forms of TB, such as lymph node TB there was limited or no evidence on the use of the shorter regimen. However, GDG members felt that the use of the shorter regimen could be considered because favourable outcomes were reported using the shorter regimen in patients with extensive disease. Subgroups excluded from the recommendation However, there were also subgroups for which there was no evidence (as they were not eligible for inclusion in the trial) and therefore the use of the shorter regimen outside the research environment is not indicated in these populations. These groups include:

• people weighing less than 40 kg;
• people with certain forms of extra-pulmonary TB (such as TB meningitis, disseminated TB, osteoarticular TB, abdominal TB);
• persons living with HIV infection with a CD4 count less than 100 cells/mm3 (NB: The trial did not include persons living with HIV infection if they had a CD4 count of less than 100 cells/mm3 and the GDG panel expressed concerns at an increased risk of relapse in this group (also because this group is at a higher risk of disseminated TB);
• children less than 12 years of age (NB: The trial aimed to recruit people aged 12 years and above. The youngest participant was 13 years of age. Therefore, no children were included in the trial. In the microbiologically-eligible population, there were 70 and 56 participants who were under 20 years of age in the rifapentine-moxifloxacin and control arms respectively); and
• pregnant, breastfeeding and postpartum women (NB: Pregnant or breast-feeding women were excluded from the study because of uncertainties about the safety of rifapentine, moxifloxacin, and pyrazinamide in these groups. Women who became pregnant while receiving study regimens were deregistered from the study and were treated according to national TB programme or local guidelines. The women continued to receive scheduled study follow-up, were classified as being on a non-study regimen, and did not receive study radiographs. Women who became pregnant while on study follow-up (but not on study treatment) continued to receive scheduled study follow-up and did not receive study radiographs. In all cases – i.e. whether pregnant during treatment or during follow up – the outcome of the pregnancy was reported on study forms).

Implementation considerations A number of implementation considerations were discussed by the GDG. These included the following: Drug susceptibility testing: The panel agreed that national TB programmes should strive for universal DST. The panel also acknowledged that universal DST is not always available but rapid DST for key medicines, including rifampicin, isoniazid and the fluoroquinolones is available and is expanding at an accelerated pace. Rapid genotypic testing for TB and rifampicin resistance is recommended by WHO as an initial test for TB and, if the same sputum sample can be tested for drug susceptibility for the fluoroquinolones and isoniazid, this can facilitate assignment of the most effective regimen. This would clearly have implications in terms of logistics, laboratory workload and cost. Balancing the desired situation of having the universal DST with reality, the panel considered that although desirable, baseline DST for fluoroquinolones would not be essential when patients with TB receive a WHO-recommended rapid molecular diagnostic test to detect rifampicin resistance. Fluoroquinolone resistance in new patients with DS-TB can reach up to 15% [1], although it is significantly lower in most settings [28-32]. In countries with high prevalence of resistance to fluoroquinolones in new patients DST for the fluoroquinolones would be highly recommended at baseline. Directly observed treatment: Patients in the trial received daily treatment that was directly observed at least five days per week. However, this may not be possible in programmatic settings. Directly observed treatment may be important in view of the pill burden and the lack of a fixeddose combination formulation, and also as a measure to prevent potential amplification of drug resistance. Current WHO recommendations support the use of directly observed treatment and also other forms of patient support and, overall, even though this regimen is a 4-month one and shorter than the current standard of care, patient support remains an important element of TB programming. Pill burden: At present, the overall pill burden will be higher for patients who will receive this 4-month regimen13 because no fixed dose combination tablet exists for the regimen and the dose of rifapentine is high (1200 mg). This may affect acceptability by patients currently, however this situation may change in future as uptake of this regimen improves, creating a demand for the regimen and its component medicines. Wider availability of rifapentine formulation of 300 mg14 may decrease the pill burden and facilitate the implementation of this new regimen until the FDC tablet becomes available. Cost of medicines: The current cost of the shorter regimen is substantially15 higher than the standard of care, mainly due to the inclusion of rifapentine. Again, this situation may change in future as uptake of the regimen improves, creating a demand for the regimen and for the medicines in it. Administration of the shorter regimen with food may present a challenge in some settings. In the trial, a flat dose of 1200 mg of rifapentine was dosed daily, with food. This was based on:

• 1)demonstration of the safety of rifapentine at 1200 mg in phase I and phase II trials; 2) demonstration that body weight does not significantly affect rifapentine clearance; 3) recognition of an effect of food in increasing rifapentine absorption [33]; and 4) modelling predictions that the target rifapentine exposure (area under the curve [AUC] of approximately 500–600 mcg*h/L) is achievable using this strategy – see the supplementary appendix to reference 13). As described in the trial’s statistical analysis plan, pharmacokinetic/pharmacodynamic modelling predicted that a rifapentine dose of 1200 mg without food would yield an AUC approximately the same as that of a rifapentine dose of 900 mg with a very high fat meal. Since the target rifapentine AUC lies somewhere between that achieved with a very high fat meal and a rifapentine dose of 900–1200 mg, the strategy proposed was a rifapentine dose of 1200 mg with a modest food requirement. The rationale was that a very high fat meal may not be feasible under trial or routine TB care conditions, whereas dosing with food may be feasible. Training of health-care workers was another implementation consideration that the panel discussed. Training will be necessary when introducing the shorter regimen into a programmatic setting. However, this is a requirement for any new programmatic intervention and the ability to shorten treatment and potentially treat more patients may offset initial training investments. Another implementation consideration discussed by the GDG concerned the choice of regimen to treat DS-TB. The GDG considered that, when choosing between the shorter 4-month regimen or the 6-month regimen, clinicians should consider eligibility criteria for the regimen and patient preference as well as local factors such as the availability of rifapentine.