Challenges of Using a Bedaquiline-Based Regimen to Treat Multidrug-Resistant Tuberculosis in Pregnancy

Abstract

The treatment of multidrug-resistant tuberculosis (MDR-TB) in pregnancy has been historically limited by the exclusion of pregnant and breastfeeding persons from large clinical studies. We follow the case of a pregnant patient successfully treated for MDR-TB in California, United States, using a bedaquiline-based regimen and discuss the growing body of evidence and evolution of recommendations in the management of this patient population. We discuss obstacles in the patient's care, including issues of regimen selection, question of breastfeeding, supratherapeutic drug levels postpartum, and medication adverse effects—all which should be considered in the management of pregnant persons. Based on our experience, we would recommend to consider monitoring drug levels during this transition to postpartum status. Further inclusion of pregnant and breastfeeding patients in MDR-TB studies will continue to bolster knowledge base and improve patient outcomes.

We describe the experience of treating multidrug-resistant tuberculosis (MDR-TB) in a pregnant patient in California, United States, in 2023. Our patient presented at 29 + 4 weeks of gestational age (WGA) with 1 month of worsening dyspnea. Her history was notable for anemia, depression, and fetal intrauterine growth restriction (IUGR) during her current pregnancy. She had immigrated from Southeast Asia several years prior and had not returned since. She denied any personal or family history of tuberculosis (TB) or any known TB exposures. Two interferon-gamma release assays several years after her immigration had been negative. She had a normal chest X-ray (CXR) at 7 + 4 WGA during an emergency department visit for hyperemesis gravidarum.

Initial laboratory tests were notable for normocytic anemia with hemoglobin 9.4 g/dL, elevated alkaline phosphatase 145 U/L, and hypoalbuminemia 2.6 g/dL. Otherwise, labs were unremarkable with white blood cell (WBC) count 7.2 cells/µL, platelets 240 cells/µL, and normal aspartate aminotransferase (AST), alanine aminotransferase (ALT), and bilirubin levels. Human immunodeficiency virus testing was negative. Her admission CXR showed bilateral pleural effusions and lower lung opacities. Pursuant chest computed tomography demonstrated multiple consolidations and multifocal nodularity suggestive of a granulomatous process. She acutely decompensated with worsening shortness of breath within 24 hours of admission, requiring bilateral chest tube placement. Vancomycin, piperacillin-tazobactam, and azithromycin were started empirically. Bronchoalveolar lavage studies were obtained with negative acid-fast bacilli (AFB) smears but a positive cartridge-based nucleic acid amplification test for TB with a mutation in rpoB in probe region E, suggestive of rifampin-resistant TB (RR-TB).

 .

Predelivery Regimen

After patient stabilization and discussions with the public health department, a treatment regimen was initiated at 30 + 5 WGA with the goal of transitioning to a bedaquiline (B), pretomanid (Pa), linezolid (L), and possibly moxifloxacin (M) combined regimen (BPaL[M]) based on the World Health Organization (WHO) consolidated guidelines and provisional Centers for Disease Control and Prevention (CDC) recommendations for MDR-TB for nonpregnant persons at that time [1, 2]. Bedaquiline was requested at the beginning of treatment; however, given challenges with obtaining the medication in the United States, it took 10 days to arrive. Pretomanid was avoided given limited antenatal safety data [3]. While first-line HRZE regimen medications (isoniazid, rifampin, pyrazinamide, and ethambutol) are considered safe in pregnancy, rifampin and isoniazid were not appropriate due to concern for MDR-TB [4]. In addition, the patient had developed transaminitis with AST and ALT rising to 212 and 224 U/L, respectively, prior to starting therapy, thought to be due to β-lactam antibiotics, which affected regimen selection. Pyrazinamide was also avoided given risk for liver toxicity. She was thus initially placed on a regimen of oral daily ethambutol (EMB) 1000 mg, linezolid 600 mg, moxifloxacin 400 mg, and cycloserine 250 mg with pyridoxine supplementation. Linezolid levels during pregnancy showed a trough of 1.46 μg/mL (goal <2 μg/mL) and 2-hour peak of 10.83 μg/mL (goal 12–24 μg/mL) [5, 6]. Her QTc was normal at 450 ms. She had been underweight prepregnancy but had a normal pregnancy weight up to 50.5 kg.

The CDC molecular detection of drug resistance testing subsequently confirmed mutations in rpoB, inhA, katG, embB, and pncA, denoting rifampin, isoniazid (inhA and katG), EMB, and pyrazinamide resistance, respectively. Molecular markers of resistance to bedaquiline, fluoroquinolones, amikacin, cefazolin, or linezolid were not detected. EMB was discontinued with the arrival of bedaquiline and the patient initiated loading on day 15 of treatment, completing 2 weeks of oral bedaquiline 400 mg daily followed by transition to 200 mg 3 times per week as per standard dosing. Her AFB sputum cultures cleared by 34 + 4 WGA, approximately 4 weeks after initiating treatment.

Postdelivery Regimen

Our patient underwent an induced vaginal delivery at 37 + 0 WGA due to fetal IUGR. Postpartum, she transitioned to the BPaLM regimen with discontinuation of cycloserine and initiation of oral pretomanid 200 mg daily. Phenotypic drug susceptibility testing confirmed resistance to rifampin, isoniazid, and pyrazinamide with susceptibility to EMB, moxifloxacin, amikacin, capreomycin, and ethionamide. Her postpartum weight dropped to 43.4 kg; however, medications in her regimen are not typically adjusted based on weight in adults. Approximately 9 weeks after initiating treatment, her labs were notable for worsening thrombocytopenia (nadir 108 platelets/µL) and decreased WBC count (nadir 2.9 cells/µL) with lymphopenia (nadir absolute lymphocytes 500 cells/µL). She also had worsening bilateral lower extremity peripheral neuropathy. A repeat linezolid trough was found to be supratherapeutic at 10.22 μg/mL with a 2-hour peak of 31.14 μg/mL. Linezolid dosing was subsequently adjusted from 600 mg daily to 3 times weekly, approximately 17 weeks into treatment. A repeat trough within 2 weeks was 0.33 μg/mL, at goal. She experienced persistent bilateral lower extremity neuropathy despite dose reduction, so fluoroquinolones were additionally held approximately 4 months into treatment. Electromyography testing and neurologic evaluation were normal.

The patient completed approximately 8 months of therapy, including a 6-month course from initiation of the BPaL(M) regimen. The full course of treatment is described in Figure 1. She had negative sputum AFB cultures at 6 and 12 months post–treatment completion. She continued to experience significant bilateral lower extremity neuropathy at 18 months posttreatment.

Figure 1.

Timeline of patient regimen denoting treatment and important considerations at each point of regimen adjustment. Created in BioRender. Giuliani, A. (2026), https://BioRender.com/flkfj6t. Abbreviations: AFB, acid-fast bacilli; BPaL(M), bedaquiline, pretomanid, linezolid, and possibly moxifloxacin; HRZE, isoniazid, rifampin, pyrazinamide, and ethambutol; IUGR, intrauterine growth restriction; NAAT, nucleic acid amplification test; RR-TB, rifampin-resistant tuberculosis; Tx, treatment; WGA, weeks of gestational age.

DISCUSSION

Predelivery Considerations

Evidence-based treatment of RR-TB and MDR-TB in pregnant patients has historically been limited by exclusion in drug studies and clinical trials. Bedaquiline was US Food and Drug Administration approved in 2012, the first new TB drug in 40 years. With the emergence of bedaquiline-based regimens, providers have been relying on small datasets and observational studies to drive care in pregnancy. Major trials including Nix-TB, ZeNix, and TB-PRACTECAL all excluded patients who were pregnant, breastfeeding, or planning to conceive, although small numbers of unintended pregnancies were captured [7–10]. In 2023, the data regarding bedaquiline-based regimens in pregnancy remained minimal. WHO and CDC guidelines were supportive of BPaL(M) for nonpregnant persons while guidance for pregnant individuals as per the Infectious Diseases Society of America (IDSA) was for a longer individualized regimen.

An observational cohort study published in 2021 helped guide our care, describing maternal treatment and pregnancy outcomes in MDR-/RR-TB including 58 pregnant persons receiving bedaquiline [11]. While multivariate analysis showed increased risk of low birthweight with bedaquiline exposure (45% vs 26%, P = .034), exposed infants also had an increased rate of normal development [11]. Other studies of patient outcomes with MDR-/RR-TB, published since our case, have admitted small populations of pregnant persons. These include the endTB multicenter observational study involving bedaquiline- and delamanid-based regimens and the BEAT Tuberculosis study, comparing bedaquiline-based regimens [12, 13]. Pretomanid has not been sufficiently studied in pregnancy and is not currently recommended [3, 10].

Regarding linezolid, case reports of use in pregnancy have not demonstrated adverse effects in the late third trimester [14]. A 2022 meta-analysis evaluated treatment outcomes of pregnant patients with MDR-TB, including about 30 treated with linezolid [15]. Treatment success was higher with linezolid, possibly confounded by use of bedaquiline and delamanid, and was primarily limited by bone marrow suppression [15]. While there has been historic concern regarding fluoroquinolones during pregnancy, a 2018 meta-analysis and a 2019 systematic review found no association with fetal malformations, preterm delivery, stillbirth, or miscarriage [16, 17]. Cycloserine, considered second-line for MDR-TB, was added as a fourth prenatal agent. While the safety profile in pregnancy is unknown, rat studies have not demonstrated teratogenic effects, and it is reserved for situations when maternal benefit outweighs potential risks [18]. It carries a 6% risk of adverse psychiatric reactions and necessitated close observation for our patient with depression [18].

Postdelivery Considerations: Breastfeeding

Breastfeeding was considered in the regimen selection. One study in 13 patients evaluated bedaquiline levels in breast milk samples compared to maternal and infant plasma [17]. Maternal plasma levels were higher than breast milk with a ratio of 14:1 [19]. Of 4 infant plasma samples, 3 non-breastfed infants demonstrated detectable but nontherapeutic levels [19]. One breastfed infant had levels similar to maternal plasma levels [19]. An animal study of pretomanid showed milk concentrations at 59%–68% of plasma levels; however, there is no information on human lactation [3]. A 2023 pharmaceutical review of MDR-TB drugs identified 3 reports of linezolid excretion in breastmilk, although at subtherapeutic amounts [20]. Levofloxacin is also excreted at subtherapeutic amounts and is not considered harmful to the infant [20]. Although there are no human studies of moxifloxacin excretion, its use is considered acceptable [21]. After an informed discussion, our patient elected not to breastfeed and to continue with the BPaLM regimen.

Postdelivery Considerations: Therapeutic Drug Monitoring

Therapeutic drug monitoring (TDM) was performed given the concern for pregnancy-related changes in absorption. A 2025 update to joint clinical practice guidelines by the American Thoracic Society, CDC, European Respiratory Society, and IDSA, published since our case, recommends linezolid levels to be obtained after 1–2 weeks of administration if able [5]. Goal levels were based on Curry International Tuberculosis Center guidelines for treatment of drug-resistant TB, although these are not specific to the pregnant population [6]. While linezolid trough and 2-hour peak levels were appropriate during pregnancy, repeat levels obtained 7 weeks after delivery due to adverse effects were supratherapeutic. Given these abnormal levels as well as persistent cytopenias and peripheral neuropathy, the dose was adjusted from daily to 3 times weekly with improvement in cytopenias. The lower prepartum levels were presumed to be related to a high volume of distribution during pregnancy, leading to supratherapeutic postpartum levels at the same dosage.

Other drug levels were subtherapeutic during pregnancy. The cycloserine 2-hour peak was 12.96 μg/mL (goal 20–35 μg/mL), prompting dose increase to 500 mg daily [6, 22]. The moxifloxacin 2-hour peak was slightly low at 2.17 μg/mL (goal 3–4 μg/mL), but the patient was maintained at her dose of 400 mg daily [20]. No other postpartum drug levels were obtained. Another case report describes similar experiences of lower linezolid and moxifloxacin concentrations during pregnancy compared to postpartum levels, underlining the need for drug monitoring during this period of rapid change [23]. Prior studies of bedaquiline-based regimens in pregnancy have not included TDM.

For her treatment duration, we ultimately decided to complete 6 months of treatment from initiation of the full BPaLM regimen given lack of evidence for a shortened course.

Postdelivery Considerations: Neuropathy

Our patient developed unremitting peripheral neuropathy despite linezolid dose reduction. Given symptom persistence, moxifloxacin was discontinued with some improvement, although not resolution, and later levofloxacin rechallenge exacerbated the neuropathy so fluoroquinolones were held for the remainder of treatment [24, 25]. While supratherapeutic linezolid was the likely culprit, fluoroquinolone administration may have worsened symptoms. The mechanism of fluoroquinolone neuropathy is unknown and the risk is low [24, 25]. One case-control study noted increased risk for each day of use with ongoing risk up to 180 days after exposure [26]. Despite regimen adjustments, our patient continued to experience neuropathy 18 months after treatment completion.

CONCLUSIONS

This case encompasses several important points in the treatment of MDR-/RR-TB in the pregnant patient. Mounting evidence suggests safety and efficacy of the use of bedaquiline- and delamanid-containing regimens in these patients, with relevant studies reviewed in Table 1 and medications in Table 2. Although, while the same class as delamanid, data on pretomanid safety in pregnancy remain limited. The BDLLfxC (bedaquiline, delamanid, linezolid, levofloxacin, and/or clofazimine if fluoroquinolone-resistant) regimen, established after this case, is considered an acceptable regimen by the WHO in treating MDR-/RR-TB in pregnancy [27]. Regimen selection should be driven by susceptibility testing, attention to drug adverse effects, and patient case characteristics. Institutions caring for large populations of birthing patients or in areas of high TB incidence should discuss proactive procurement of bedaquiline to facilitate timely treatment. TDM may be considered in pregnant patients in high-resource settings, particularly in transition to the postpartum period. Linezolid levels are routinely recommended by joint guidelines given variability of concentrations, even at normal doses, and attention may be paid to cycloserine, as increased neuropsychiatric effects are seen at higher serum concentrations [18, 22, 28]. Although several drugs may be excreted in breast milk, there are limited data on effects in the infant; most can be continued with close surveillance. Ongoing inclusion of pregnant and lactating patients in MDR-TB treatment studies, including monitoring prepartum and postpartum drug levels, is critical to determine best practices and provide evidence-based care.

Table 1.

Studies of Multidrug-Resistant Tuberculosis Treatment With Attention to Inclusion of Pregnant and Breastfeeding Patients

Study	Purpose	Pregnancy Outcomes
Observational (Loveday et al) [11]	Observational cohort study describing maternal treatment and pregnancy outcomes in MDR-TB/RR-TB	Included 108 pregnant patients: 58 patients received bedaquiline Increased risk of low birthweight (45% vs 26%, P = .034) with bedaquiline exposure but also higher rate of normal development (88% exposed vs 82% not exposed) Levofloxacin independently associated with risk of low birthweight
endTB [12]	Compared oral regimens for treatment of MDR-TB/RR-TB, including 9-mo bedaquiline-containing regimens	Included 43 pregnant patients (31 continued pregnancies): 98% successful treatment 81% live births, 13% unknown outcome, 2% spontaneous abortion in first trimester (of continued) 32% low birthweight infants (of continued and known birthweight) No birth malformations
BEAT Tuberculosis [13]	Demonstrated noninferiority of 6-mo course BDLLfxC vs 9-mo or more course of standard of care bedaquiline-based regimen	Included 10 pregnant patients 100% (n = 6) successful treatment in control group 75% (n = 3) successful treatment in study group 100% live births One premature delivery Included breastfeeding patients (not further delineated)

Abbreviations: BDLLfxC, bedaquiline, delamanid, linezolid, levofloxacin, and/or clofazimine if fluoroquinolone-resistant; MDR-TB, multidrug-resistant tuberculosis; RR-TB, rifampin-resistant tuberculosis.

Table 2.

Multidrug-Resistant Tuberculosis Medications and Implications in Pregnancy and Breastfeeding

Drug	Pregnancy Category	Pregnancy	Fetal Outcomes	Breastfeeding
Bedaquiline	Category B	Efficacy and safety demonstrated in endTB [12] and BEAT Tuberculosis [13] trials No teratogenicity	Increased risk of low birthweight [11] No congenital malformations reported [11–13] Normal development, improved outcome compared to nonexposed [11]	Excreted in milk [19] Monitor infant
Delamanid (not readily available in United States)	Category C	Efficacy and safety demonstrated in endTB [12] and BEAT Tuberculosis [13] trials No adverse outcomes in cohort studies	No adverse neonatal outcomes in cohort studies [12, 13]	Excreted in milk [29] Monitor infant
Pretomanid	Category C	Insufficient studies, not currently recommended although is of same class as delamanid [3, 10]	No teratogenicity in animal studies at higher doses [3]	No human lactation data [3]
Linezolid	Category C	Efficacy and safety demonstrated in endTB [11] and BEAT Tuberculosis [12] trials No adverse effects reported in late third trimester [14, 15] Use limited by cytopenias [15]	No increased risk miscarriage or major birth defects [16]	Excreted in milk, subtherapeutic [20] Monitor infant
Fluoroquinolone	Category C	No association with adverse pregnancy outcomes in meta-analysis and systematic review [16, 17]	No association with fetal malformations, preterm delivery, stillbirth, or miscarriage [16, 17]	Excreted in milk, subtherapeutic [20, 21] Monitor infant
Cycloserine	Category C	Risk of psychiatric adverse events (as in general population) [18]	No clear teratogenic effects in animal studies [18]	No human lactation data