Safety and Effectiveness of 3 Novel All-Oral Shortened Regimens for Rifampicin- or Multidrug-Resistant Tuberculosis in Kazakhstan Open Access

Abstract

Background

In 2019, the World Health Organization called for operational research on all-oral shortened regimens for multidrug- and rifampicin-resistant tuberculosis (MDR/RR-TB). We report safety and effectiveness of three 9-month all-oral regimens containing bedaquiline (Bdq), linezolid (Lzd), and levofloxacin (Lfx) and reinforced with cycloserine (Cs) and clofazimine (Cfz), delamanid (Dlm) and pyrazinamide (Z), or Dlm and Cfz.

Methods

We conducted a prospective cohort study of patients initiating treatment for pulmonary MDR/RR-TB under operational research conditions at public health facilities in Kazakhstan. Participants were screened monthly for adverse events. Participants with baseline resistance were excluded from the study and treated with a longer regimen. We analyzed clinically relevant adverse events of special interest in all participants and sputum culture conversion and end-of-treatment outcomes among individuals who were not excluded.

Results

Of 510 participants, 41% were women, the median age was 37 years (25th–75th percentile: 28–49), 18% had a body mass index <18.5 kg/m², and 51% had cavitary disease. A total of 399 (78%) initiated Bdq-Lzd-Lfx-Cs-Cfz, 83 (16%) started Bdq-Lzd-Lfx-Dlm-Z, and 28 (5%) initiated Bdq-Lzd-Lfx-Dlm-Cfz. Fifty-eight individuals (11%) were excluded from the study, most commonly due to identification of baseline drug resistance (n = 52; 90%). Among the remaining 452 participants, treatment success frequencies were 92% (95% CI: 89–95%), 89% (95% CI: 80–94%), and 100% (95% CI: 86–100%) for regimens with Cs/Cfz, Dlm/Z, and Dlm/Cfz, respectively. Clinically relevant adverse events of special interest were uncommon.

Conclusions

All regimens demonstrated excellent safety and effectiveness, expanding the potential treatment options for patients, providers, and programs.

Multidrug- or rifampicin-resistant tuberculosis (MDR/RR-TB) newly afflicts an estimated 410 000 people annually [1]. Historically, treatment success rates have been low, at 63% globally [1]. Reasons for this included the long treatment duration (18–24 months) and the use of toxic drugs with limited efficacy. An important strategy to improve the treatment outcomes for MDR/RR-TB has been through research on shortened (6- to 9-month) all-oral regimens. Pivotal clinical trials, including standard treatment regimen of anti-tuberculosis drugs for patients with MDR-TB (STREAM) Stage 2, TB-pragmatic clinical trial for a more effective, concise and less toxic regimen (PRACTECAL), expand new drug markets for tuberculosis (endTB), and Nix-TB [2–5], demonstrated the tremendous potential of all-oral shortened regimens in improving treatment outcomes, with success rates ranging from 83% to 90%. The success of early trials of all-oral shortened regimens led the World Health Organization (WHO) in 2020 to issue guidance that some regimens could be used under operational research conditions [6]. In 2022, WHO reinforced that additional research was needed on the effectiveness and safety of variants of all-oral shortened (6- or 9-month) MDR/RR-TB regimens, particularly those including bedaquiline (Bdq) [7].

STEM-TB (NCT05871489) is a multi-country prospective cohort study with a primary objective of reporting the effectiveness and safety of novel, all-oral, shortened MDR/RR-TB regimens when implemented under operational research conditions in Lesotho, Peru, and Kazakhstan. Here, we describe sputum culture conversion, end-of-treatment outcomes, and clinically relevant adverse events of special interest under three 9-month all-oral regimens containing a core of 3 WHO Group A drugs—Bdq, linezolid (Lzd), and levofloxacin (Lfx)—in combination with 2 additional TB drugs—when implemented under operational research conditions in Kazakhstan.

METHODS

Study Setting

Kazakhstan is a central Asian country with a low human immunodeficiency virus (HIV) prevalence (0.3%) [8] and one of the highest burdens of MDR/RR-TB in the world. Twenty-seven percent of individuals newly diagnosed with TB and 44% of individuals requiring TB retreatment have MDR/RR-TB [9]. End-of-treatment success rates among patients receiving longer regimens containing new and repurposed drugs for MDR/RR-TB in Kazakhstan are high compared with many settings: between 2016 and 2018, clinician-reported success rates in the endTB observational study were 93% (95% confidence interval [CI]: 84–97%) in patients with no prior TB treatment history or first-line treatment only and 87% (95% CI: 84–90%) in patients previously treated with second-line drugs (M. F. Franke, unpublished data 2024).

Study Design and Population

Patients aged 18 years and older with bacteriologically confirmed MDR/RR pulmonary TB were eligible for the operational research study on all-oral 9-month treatments if they consented to study participation, including study follow-up. Because the aims of this study were descriptive, rather than comparative, the sample size was determined by the number of consecutive individuals initiating treatment at a participating health facility during the study period. This analysis included patients enrolled at 5 public health facilities in Kazakhstan from 22 September 2020 through 30 September 2022. Patients were excluded from the operational research if unable to take oral medication, if taking any medications contraindicated with drugs in the 9-month regimens, or if they had any of the following: a known allergy or hypersensitivity to a drug in the regimen; any prior documented resistance to a fluoroquinolone (or other drugs for which there was a reliable drug susceptibility test [DST]); previous treatment with 1 or more second-line drugs in the regimen (>1 month); extrapulmonary TB without pulmonary TB; severe or intractable extrapulmonary TB (eg, TB meningitis or miliary TB); a baseline QTc (QT interval corrected by the Fridericia method) of 500 ms or more on electrocardiogram (ECG), despite correction of serum electrolytes; a baseline aspartate aminotransferase (AST) or alanine aminotransferase (ALT) greater than 3 times the upper limit of normal; or a baseline creatinine clearance below 30 mL/minute per 1.73 m² body surface area.

Standardized All-Oral Shorter Regimens

Three 9-month all-oral regimens were used. All regimens had a core consisting of Bdq, Lzd, and Lfx (ie, 3 WHO Group A drugs), and were reinforced with 2 additional drugs: cycloserine (Cs) and clofazamine (Cfz), delamanid (Dlm) and pyrazinamide (Z), or Dlm and Cfz. Clinicians selected a regimen for each patient according to the guiding principles and considerations in Supplementary Table 1. The intention was to use only drugs that were “likely effective” in the individual patient. Likely effectiveness was based on confirmed susceptibility of the infecting strain of Mycobacterium tuberculosis or no prior use of the medicine for more than 1 month in the absence of a DST. The 1 exception was pyrazinamide, which could be used regardless of its likely effectiveness. Participants who, after enrollment, were identified as having a strain of M. tuberculosis resistant to fluoroquinolones, Bdq, Lzd, Dlm, and/or Cfz, at the time treatment initiation (ie, from a pretreatment sputum sample or from a sputum sample within the first 30 days of treatment) were excluded from the study and treated with an 18–20-month individualized regimen. In addition, individuals who had 2 or more drugs permanently discontinued were assigned a treatment outcome, removed from the study regimen, and treated per national guidelines, typically with an 18–20-month regimen. Treatments could be extended up to a maximum total duration of 12 months at the clinician's discretion due to a slow clinical response.

Study Procedures

Treatment delivery and clinical monitoring occurred as part of routine care. Baseline assessments included a medical history, a comprehensive clinical examination, a pregnancy test, HIV counseling and testing, sputum and blood collection for laboratory tests, chest X-ray, and ECG. Sputum-based analyses included Xpert MTB/RIF (Cepheid) (or other molecular diagnostic resistance testing), sputum smear microscopy and culture, and culture-based first- and limited second-line DST. Sputum cultures were grown on both solid (Lowenstein-Jensen) and liquid (BACTEC MGIT, Becton Dickinson) media. Second-line DSTs were conducted on liquid medium (BACTEC MGIT, Becton Dickinson) at the regional bacteriological laboratories, which are certified as reference laboratories. The DSTs for Bdq, Dlm, Lzd, and Cfz were introduced and scaled in Kazakhstan between 2019 and 2022, depending on the drug. Other laboratory tests comprised complete blood count and hemoglobin, creatinine/urea, electrolytes, albumin, liver function tests including lipase, and hepatitis B and C serology.

Monthly follow-up assessments throughout treatment included a review of signs and symptoms, clinical examination (including screening for visual acuity, color blindness, peripheral neuropathy), sputum smear microscopy and culture (>75% of on-treatment follow-up cultures were grown on liquid medium), monthly hemotology and biochemistry, and ECG (at 2 weeks and monthly thereafter while on Bdq).

Adverse Events Reporting

Reported adverse event outcomes consisted of a subgroup of all possible adverse events and included hepatotoxicity, peripheral neuropathy, optic neuritis, QTc prolongation, myelosuppression, seizure, and psychosis. These adverse events of special interest (AESIs) were predefined as clinically relevant events [10] that may be caused by using medications included in the shortened regimens. Optic neuritis was screened using a visual acuity exam and the Ishihara Color Blindness test. Patients were asked to report any changes in vision during their participation in the study. Those who reported vision changes and/or experienced deteriorating visual acuity and/or Ishihara Color Blindness results were referred to an ophthalmologist and Lzd was stopped. Peripheral neuropathy was assessed via the Brief Peripheral Neuropathy Screen [11]. Table 1 shows the severity grade at which an AESI was considered to be clinically relevant. In addition to AESIs, we recorded all serious adverse events [12].

Statistical Analysis

We performed 2 analyses: one to assess the safety of the shortened oral regimens and the other to evaluate effectiveness. Safety analyses were conducted among all participants and effectiveness analyses were conducted among individuals who were not excluded from the study. Safety analyses were restricted to adverse events and person-time corresponding to time on a study regimen. If an individual discontinued a study regimen (eg, due to baseline resistance) and initiated a longer individualized regimen, adverse events and person-time corresponding to the former, but not the latter, were analyzed. We summarized categorical variables with frequencies, percentages, and continuous variables using medians and interquartile ranges (IQRs). For each clinically relevant AESI, we calculated the percentage of participants experiencing at least 1 episode, the median time to the first occurrence, and the incidence per 1000 person-months of treatment with its 95% CI. We calculated person-months of exposure from the start of treatment until the event or the end of treatment. We conducted sensitivity analyses in which we restricted safety analyses to those who were included in effectiveness analyses.

Effectiveness outcomes included the frequency of and time to sputum culture conversion and end-of-treatment outcome. Sputum culture conversion was defined among those with a positive baseline sputum culture as 2 consecutive negative cultures, spaced at least 7 days apart. Baseline cultures were those occurring from 90 days before to 7 days after treatment initiation. Although defining baseline culture status from a post-treatment sample can introduce bias when early losses to follow-up or death are frequent [13, 14], there were no early deaths or losses within the first week of treatment in this cohort. End-of-treatment outcomes were determined by clinicians, according to national treatment guidelines, which were informed by 2013 WHO treatment outcome definitions with 1 notable exception: baseline sputum smear, rather than culture, determined whether an outcome of cure could be established [15, 16]. Individuals assigned end-of-treatment outcomes of “cure” or “treatment completion” were considered to have successfully completed treatment. Loss-to-follow-up was defined as a treatment interruption of 28 days or more.

Confidence intervals were calculated using the Fisher's exact method. All data were analyzed in SAS version 9.4 (SAS Institute, Cary, NC). The Supplementary Material contains additional information on the study team, adherence support, adverse event grading, data management, and end-of-treatment outcome definitions.

Research Ethics

This study was approved by the Ethics Committee of the National Scientific Center for Phthisiopulmonology of the Ministry of Health of Kazakhstan and Harvard Longwood Campus Institutional Review Board. All participants signed written informed consent.

RESULTS

Overview

The 510 participants enrolled in Kazakhstan during the study period comprised the cohort for safety analyses (Supplementary Figure 1). A total of 399 (78.2%) initiated Bdq-Lzd-Lfx-Cs-Cfz, 83 (16.3%) started Bdq-Lzd-Lfx-Dlm-Z, and 28 (5.5%) initiated Bdq-Lzd-Lfx-Dlm-Cfz. Of the 510 participants included, 41.2% were women, the median age was 37 years (25th–75th percentile: 28–49 years), 51.3% had cavitary disease, and 18.2% had a body mass index less than 18.5 kg/m² (Table 2). All participants had bacteriologically confirmed MDR/RR-TB through genotypic or phenotypic DSTs. Four hundred and twenty-one participants (82.5%) had a fluoroquinolone DST at baseline (ie, from a sample collected any time prior to treatment initiation up through the first 30 days of treatment) (Supplementary Table 2).

Fifty-eight individuals (11.4%) were excluded from the study after treatment initiation, and therefore excluded from effectiveness analyses. This occurred most commonly due to identification of baseline resistance to a fluoroquinolone, Lzd, Bdq, and/or Cfz (n = 52; 89.7%) (Supplementary Figure 2). The median time from treatment initiation to exclusion was 60 days (25th–75th percentile: 34–102 days).

Safety

Clinically relevant AESIs were infrequent (Tables 3 and 4) and only 1 clinically relevant AESI became a severe adverse event. Peripheral neuropathy was experienced by 18 participants (3.5%), resulting in a permanent drug discontinuation in 3 participants (16.7%). Myelosuppression was the second most common AESI, experienced by 13 participants (2.5%). QTc prolongation was notably infrequent, experienced by only 2 participants (0.4%), with no permanent drug cessation. Four participants developed psychosis, all of whom were receiving the Cs-containing regimen. Additionally, the Cs-containing regimen was associated with 5 out of the 6 reported cases of seizures. While clinically relevant hepatotoxicity was only experienced by 1 participant, lower grades of hepatotoxicity that did not meet the clinically relevant threshold occurred more frequently, with 96 (18.8%) participants experiencing grade 1 hepatotoxicity and 1 participant experiencing grade 2 hepatotoxicity (0.20%). Safety analyses restricted to individuals included in effectiveness analyses were similar (Supplementary Tables 3 and 4). Adverse event monitoring occurred regularly (Supplementary Table 5).

Effectiveness

Two hundred and eighty-eight of 452 participants (63.7%) had a positive baseline culture, of whom 287 (99.7%) achieved sputum culture conversion within 5 months. The median number of monthly on-treatment follow-up cultures was 9 (25th–75th percentile: 7–9). The median time to conversion among those whose sputum converted was 1.0 month (25th–75th percentile: 0.9–1.8 months). Plots of time to culture conversion are shown in Supplementary Figure 2.

The overall frequency of treatment success was 92.0% (95% CI: 89.2–94.2%), with frequencies of 92.1% (95% CI: 88.9–94.5%), 89.0% (95% CI: 79.8–94.3%), and 100% (95% CI: 85.7–100%) for regimens with Cs/Cfz, Dlm/Z, and Dlm/Cfz, respectively (Table 5). The median treatment duration was 8.9 months (25th–75th percentile: 8.9–9.0 months; range: 0.7 to 12.7 months). A small fraction of participants (n = 12; 2.7%) were treated beyond 10 months.

There were 7 deaths (1.5%) and 6 losses to follow-up (1.3%) and treatment failed in 23 participants (5.1%). Of participants in whom treatment failed, the reason was drug intolerance in 5 and a slow clinical response and/or positive cultures at or after 4 months of treatment in 16 participants. Two additional participants who lacked a pretreatment fluoroquinolone DST had a resistant result from sputum collected after 1 month of treatment (at 60 days and 90 days) and were changed to a longer regimen. Because we could not distinguish between baseline and acquired resistance in these participants, we conservatively classified treatment as having failed.

DISCUSSION

Outcomes this cohort of patients receiving 1 of 3 all-oral shortened regimens were excellent, with some of the highest end-of-treatment treatment success rates reported to date in programmatic care settings. The 3 regimens performed comparably to 2 alternative shortened all-oral regimens recommended by WHO in 2022: the 6-month bedaquiline, pretomanid, linezolid, moxifloxacin (BPaLM) regimen and the 9-month, 7-drug all-oral regimen [3, 4]. These results also compare well with results from other, mostly small, cohort studies from low-HIV-prevalence settings on all-oral shortened regimens with a core of 2 or 3e Group A drugs [17–19].

Clinically relevant AESIs were infrequent This may be due to the shorter duration of treatment and the extensive experience of the Kazakh physicians in managing adverse events. While possible that adverse events were underreported, monitoring was consistent and rigorous and there was a low incidence of adverse events diagnosed through objective methods (ie, laboratory testing). Furthermore, permanent discontinuation of drugs and loss to follow-up were rare, supporting that toxicity was infrequent and/or did not affect tolerability. Linezolid toxicity was uncommon; however, among the small percentage of patients who experienced peripheral neuropathy, Lzd was discontinued in 16.7% of patients. All psychosis and the majority of seizure cases were reported in the Cs-containing regimen. Cycloserine is known to be challenging in terms of tolerance and central nervous system adverse events. While the frequency of Cs events was low in this cohort compared with others [21–24], Cs was discontinued due to an adverse event in 3.4% of participants receiving it, more commonly than any other drug. Nonetheless, the Cfz/Cs-containing regimen was often used preferentially due to the cost of Dlm, which was considerably more expensive than other drugs in the regimen and was used when available through a manufacturer donation. Because Dlm offers advantages in terms of tolerability in adults, a Dlm-containing regimen may be preferred over Cs-containing regimens by patients—and by national programs if the price of Dlm can be reduced.

A possible benefit of the 5-drug regimens used in Kazakhstan is that if Lzd dosage needs to be reduced or a drug needs to be suspended altogether, the regimen will still contain 4 drugs, likely sufficient to protect against acquisition of resistance. This contrasts with bedaquiline, pretomanid, linezolid (BPaL) and BPaLM, where the cessation of Lzd results in a 2- or 3-drug regimen, respectively. On the other hand, a 5-drug regimen could result in a higher pill burden and added toxicity overall. The advantages and disadvantages of promising shortened regimens could be evaluated via pooled analyses of existing data and large prospective studies; understanding patient preferences is also critical to treatment selection.

The STEM-TB observational study was designed to be an “all-comers” cohort that enrolled all consenting eligible individuals without additional selection; however, by restricting the cohort to patients without prior treatment with the second-line drugs in the regimen, the patient population was likely a healthier subset of all of those requiring MDR/RR-TB treatment. Individuals previously treated with second-line drugs tend to have more advanced disease. Kazakh participants in the endTB observational study [20], another “all-comers” cohort that enrolled patients regardless of treatment history, were largely similar to patients enrolled in this STEM-TB cohort, except that they more frequently had a baseline body mass index <18.5 kg/m² and bilateral disease, cavitary disease, or fibrosis on chest radiograph (Supplementary Table 7). Nonetheless, one-quarter of participants enrolled in the STEM-TB cohort had bilateral lung disease and over half had cavities on baseline X-ray, suggesting that physicians were willing to prescribe the shortened regimens for participants with advanced disease. Notably, previous research has suggested the need for extended treatment in patients with cavitary disease and high smear grades, making this, as well as quantifying the risk of relapse, a crucial area for future investigation [25, 26]. Access to second-line DSTs, and particularly to Bdq, Dlm, Lzd, and Cfz DSTs, increased throughout the study period. Ensuring access to these DSTs will be a critical element of scale-up of all-oral shortened regimens. Finally, it is possible that the additional resources for operational research facilitated data quality and adherence to clinical monitoring procedures, contributing to the overall quality of patient care.

CONCLUSIONS

Three all-oral shortened regimens implemented under operational research conditions demonstrated excellent safety and effectiveness in individuals without prior exposure to the second-line TB drugs included in their regimen. These regimens may be suitable choices for use in routine programmatic care, thereby broadening the array of treatment options for patients, healthcare providers, and programs on a global scale. Rigorous monitoring was feasible and clinically relevant adverse events were rare. While we cannot rule out underreporting, toxicity and intolerance did not have a substantial impact on the ability to complete treatment successfully. Research into optimal monitoring strategies and the best ways to communicate the pros and cons of each regimen and incorporate patient preference will be critical to achieving a patient-centered treatment experience. Moreover, continued observational research into all-oral shortened regimens remains imperative, even for those regimens that have shown success in randomized trials—to understand real-world effectiveness, relative tolerability, post-treatment outcomes, and whether certain subgroups prefer, or derive, greater benefit from particular regimens.

Supplementary Data

Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Author contributions. Study design: K. J. S., M. L. R., M. F. F., C. D. M., M. R., Y. A. Analysis design: L. T., M. F. F. Study implementation: M. R., G. B., J. S., G. E., S. B., N. A., E. O., A. C. Data management: A. L., L. T., N. A., M. R. Statistical analyses: L. T., A. L., M. F. F. Data interpretation: all authors. First draft of the manuscript: M. R., M. F. F., M. L. R., Y. A. Critical review of the manuscript: all authors. Approved final version of the manuscript: all authors.

Acknowledgments. The authors are indebted to the patients who participated in the STEM-TB observational study and to the clinicians and program staff members of participating health facilities.

Disclaimer. The funders had no role in the conceptualization, analysis, or presentation of findings of this study. The companies from which drug donations were received did not have any role in study design, data analyses, data interpretation, or manuscript writing.

Data availability. To researchers with a methodologically sound proposal, pseudo-anonymized individual participant data that underlie the results reported in the text and tables of this article and a data dictionary may be made available upon request to the corresponding author, and execution of a data sharing agreement or alternate means that allows assurance that principles of Kazakhstan regulations will be met.

Financial support. This work was supported by the Global Fund to Fight AIDS, Tuberculosis, and Malaria and the National Institute of Allergy and Infectious Diseases (grant number R01AI46095; to M. R., A. L., N. A., E. O., L. T., M. L. R., K. J. S., Y. A., C. D. M., and M. F. F.). Donations of delamanid were provided to Partners In Health by Otsuka through an Investigator Initiated Award to K. J. S.

References

Author notes