Emergence and characterisation of bedaquiline, pretomanid and linezolid resistance in recurrent rifampicin-resistant tuberculosis patient: A case report

Abstract

Background

Bedaquiline (BDQ), pretomanid (Pa) and linezolid (LZD)-based (BPaL) regimens have transformed rifampicin-resistant tuberculosis (RR-TB) treatment. However, resistance to these drugs is increasingly reported, while longitudinal data integrating clinical, phenotypic and genomic evolution remain limited.

Case presentation

We report a fatal case of recurrent RR-TB in a 40-year-old South African woman with HIV infection who received BPaL therapy. Following treatment completion, recurrent disease was complicated by pregnancy, treatment interruption and loss to follow-up. She subsequently re-presented with advanced cavitary pulmonary tuberculosis and respiratory failure. Despite re-initiation of BPaL, she deteriorated and died from septic shock. Routine drug susceptibility testing, available posthumously, confirmed resistance to BDQ, Pa and LZD. Seven sequential Mycobacterium tuberculosis isolates collected during treatment underwent minimum inhibitory concentration (MIC) testing and whole-genome sequencing. All demonstrated phenotypic resistance to the three BPaL drugs. Serial genomic analysis identified progressive Rv0678 diversification associated with a two-fold increase in BDQ MIC, while resistance-associated rrl and fbiC mutations remained comparatively stable despite persistently elevated LZD and Pa MICs.

Conclusions

This case demonstrates the interaction of treatment interruption, delayed resistance detection and continued within-host bacterial adaptation during BPaL therapy. Longitudinal phenotypic and genomic findings revealed distinct resistance trajectories among BPaL drugs, highlighting the importance of treatment retention, timely susceptibility testing and genomic surveillance for preserving BPaL effectiveness.

Introduction

The introduction of bedaquiline (BDQ), pretomanid (Pa) and linezolid (LZD)-containing regimens has transformed the management of rifampicin-resistant / multidrug-resistant tuberculosis (RR/MDR-TB) [1], [2]. Clinical trials, including Nix-TB, ZeNix and TB-PRACTECAL, demonstrated high treatment success rates with shorter all-oral BPaL-based regimens, leading to their rapid incorporation into World Health Organization (WHO) recommendations and national tuberculosis programmes [3], [4], [5]. These regimens have substantially improved treatment outcomes while reducing reliance on prolonged injectable therapies and extended treatment durations [1], [2], [6].

Despite these advances, the increasing programmatic use of BPaL has raised concerns regarding the emergence and amplification of resistance to BDQ, Pa and LZD [7], [8], [9], [10], [11], [12], [13]. Reports describing resistance to these newer agents remain relatively uncommon but are increasing, particularly in patients with prolonged treatment, treatment interruption, cavitary disease and previous exposure to second-line anti-tuberculosis drugs. Resistance to BDQ has most frequently been associated with mutations in Rv0678 and, less commonly, atpE, whereas resistance to Pa is primarily linked to mutations affecting the F420 cofactor biosynthesis pathway, including ddn , fbiA, fbiB, fbiC and fgd1 [7], [8], [9], [10], [11], [14]. Linezolid resistance is most commonly associated with mutations in the rrl and rplC genes [9], [11], [15], [16]. Although these resistance mechanisms are increasingly recognised, the temporal relationship between phenotypic resistance, genomic evolution and clinical progression remains poorly understood.

Most available evidence describing resistance to BPaL drugs originates from clinical trials, cross-sectional surveillance studies or isolated reports documenting resistance at a single time point [3], [4], [5]. Consequently, there is limited understanding of how resistance evolves within individual patients during treatment or how patient-related factors, health-system constraints and bacterial adaptation interact to influence clinical outcomes [17], [18]. Treatment-emergent resistance to individual BPaL components has been described in clinical cohorts and, more recently, in individual case reports. A recent case from Zambia documented acquired BDQ and LZD resistance during BPaL treatment following treatment interruption and delayed drug susceptibility testing, while a report from the United States described resistance involving all three BPaL components following treatment with novel short-course regimens [19], [20]. However, detailed longitudinal case-level investigations integrating the clinical course with serial phenotypic susceptibility measurements and whole-genome sequencing of multiple sequential isolates during BPaL treatment remain limited. Such analyses are important for understanding how established resistance subsequently evolves within an individual patient during treatment.

We report a fatal case of recurrent RR-TB in South Africa characterised by persistent phenotypic resistance to BDQ, Pa and LZD and progressive within-host genomic diversification during BPaL therapy. By integrating longitudinal clinical observations with serial minimum inhibitory concentration (MIC) measurements and whole-genome sequencing (WGS) of seven sequential Mycobacterium tuberculosis isolates, this report provides a detailed illustration of resistance evolution within a single patient. Beyond documenting treatment failure, the case highlights how treatment interruption, pregnancy-related loss to follow-up, delayed recognition of resistance and continued bacterial adaptation may converge to compromise treatment outcomes, providing clinically relevant insights that complement population-level studies of resistance to newer anti-tuberculosis agents.

Case presentation

A 40-year-old South African female presented with a complex history of recurrent rifampicin-resistant tuberculosis (RR-TB). She was HIV positive since 2012 on antiretroviral therapy (ART) (tenofovir-lamivudine-dolutegravir). Her latest (June 2025) HIV viral load was < 50 RNA copies/mL, with CD4 + cell count of 228 cells/µL.

Fig. 1 provides the patient’s clinical course. The patient was initially diagnosed with pulmonary RR-TB in late 2022. Baseline drug susceptibility testing demonstrated susceptibility to isoniazid (INH) and fluoroquinolones (FLQs). Despite treatment, she remained bacteriologically positive over the following months, and repeat investigations in August 2023 demonstrated persistent smear-positive pulmonary tuberculosis (smear grading 2 +) with newly acquired FLQ resistance while phenotypic drug susceptibility testing confirmed susceptibility to both bedaquiline (BDQ) and linezolid (LZD). She was subsequently referred to a specialised multidrug-resistant tuberculosis (MDR-TB) hospital in November 2023 for further management.

Fig. 1.

Clinical timeline showing treatment episodes, longitudinal research sampling, phenotypic and genomic resistance findings, and availability of routine drug susceptibility testing results.

The patient commenced an eight-month BPaL regimen consisting of BDQ, Pa and LZD in December 2023 and completed treatment in July 2024. She was classified as having completed treatment; however, approximately three months later, in October 2024, she experienced microbiologically confirmed recurrent disease with fluoroquinolone-resistant tuberculosis and was re-admitted to the MDR-TB hospital for pre-XDR TB treatment

During her treatment, the patient became pregnant but did not disclose the pregnancy to healthcare providers at the MDR-TB hospital. In August 2025 she defaulted follow-up appointments and informed healthcare workers at her local health care facility that she would only return for treatment after delivery. Attempts to trace her through hospital staff and community-based caregivers were unsuccessful, and she was subsequently classified as lost to follow-up.

In December 2025, shortly after delivery, the patient returned to the MDR-TB hospital with her newborn infant. She was admitted directly to the extensively drug-resistant tuberculosis (XDR-TB) ward following referral from a local healthcare facility. At presentation, she reported progressive productive cough, marked weight loss, dyspnoea, chest tightness, hoarseness of voice and poor appetite. Chest radiography performed on admission demonstrated extensive bilateral patchy opacities involving the mid and lower lung zones, with greater involvement of the left lung. Multiple cavitary lucencies were evident within the areas of consolidation, consistent with advanced cavitary pulmonary tuberculosis. No obvious pleural effusion or pneumothorax was identified. These radiographic findings were in keeping with severe bilateral pulmonary disease and correlated with the patient's respiratory compromise on presentation (Fig. 2). Clinical examination demonstrated hypoxaemia (oxygen saturation 87% on room air), tachypnoea and bilateral coarse crepitations consistent with severe pulmonary disease requiring supplemental oxygen. Due to maternal infectiousness, the neonate was managed as a tuberculosis contact and isoniazid and levofloxacin was initiated as preventive therapy.

Fig. 2.

Chest radiograph obtained during the patient's final admission demonstrating extensive bilateral patchy opacities with multiple cavitary lesions, more pronounced within the left lung, consistent with advanced cavitary pulmonary tuberculosis.

Following admission on 9 December 2025, sputum was collected and submitted for routine phenotypic drug susceptibility testing on the same day; therefore, there was no documented delay between recognition of the need for repeat microbiological assessment and specimen collection during this admission. Treatment was initiated according to national programmatic management of drug-resistant tuberculosis (Table 1). Based on the patient’s previous treatment history and the absence of confirmed resistance to BDQ, Pa and LZD at that time, she was commenced on a BPaL regimen. Routine phenotypic DST for the BPaL drugs remained pending throughout her subsequent clinical deterioration and was unavailable when she demised on 8 January 2026, 30 days after specimen collection. The DST result became available in February 2026, approximately two months after sampling, and confirmed resistance to BDQ, Pa and LZD (Table 2).

Table 1.

Treatment regimen and supportive management during the patient's final admission.

Drug	Dose	Purpose
Bedaquiline	400 mg daily for 14 days, then 200 mg daily	Anti-TB
Pretomanid	200 mg daily	Anti-TB
Linezolid	600 mg daily	Anti-TB
Bactrim	960 mg daily	Opportunistic infection prophylaxis
Oxygen therapy	Nasal prongs	Respiratory support
Vasopressor support	Adrenaline infusion	Septic shock management
Electrolyte replacement	Potassium supplementation	Correction of hypokalaemia

Table 2.

Selected laboratory findings during the final admission.

Parameter	12 Dec 2025	29 Dec 2025	8 Jan 2026
Sodium (mmol/L)	131	134	141
Potassium (mmol/L)	3.8	1.8	2.7
Urea (mmol/L)	4.6	9.7	10.3
Creatinine (µmol/L)	32	56	74
eGFR (mL/min/1.73 m²)	135	112	87
Haemoglobin (g/dL)	10.8	10.5	9.4
White cell count (×10⁹/L)	7.01	12.71	4.83
Platelets (×10⁹/L)	251	288	288
Albumin (g/L)	—	21	—
CD4 count (cells/µL)	228	—	—
HIV viral load	< 50 copies/mL	—	—

Initially, the patient remained clinically stable. During month 2 of her second BPaL treatment episode (BPaL-2, M2; first week of January 2026), the patient developed progressive clinical deterioration with persistent hypotension, worsening respiratory distress and progressive hypoxaemia. Despite aggressive supportive management, she developed septic shock secondary to advanced drug-resistant tuberculosis and sadly, she demised.

Several weeks after the patient demised, routine phenotypic drug susceptibility testing confirmed resistance to BDQ, Pa and LZD. To further investigate the evolution of resistance, seven sequential research isolates collected during the patient's treatment course underwent WGS and longitudinal phenotypic susceptibility testing. The findings are presented below:

Microbiological and genomic investigations

As part of the longitudinal research study, seven sequential Mycobacterium tuberculosis isolates collected between October 2024 and July 2025 during the individualized pre-XDR-TB treatment episode were analysed by WGS and MIC testing for BDQ, Pa and LZD. The first study isolate was collected in October 2024 during [month 1/baseline] of this treatment episode, followed by isolates collected in December 2024, January 2025, March 2025, May 2025, June 2025 and July 2025 ([corresponding treatment months]). These analyses were performed for research purposes and the results were not available to clinicians for treatment decision-making. No corresponding WGS data were available from the patient's initial diagnostic isolate or before her first BPaL treatment episode (December 2023–July 2024). Consequently, although earlier routine phenotypic DST in August 2023 documented susceptibility to BDQ and LZD, the precise timing of acquisition of the resistance-associated mutations subsequently identified by WGS could not be determined.

All sequenced isolates belonged to the East Asian (Beijing) lineage (Lineage 2.2.1). The first research isolate, collected in October 2024 during month 1 of the individualized pre-XDR-TB treatment episode (TE2, M1), already harboured resistance-associated variants in Rv0678 (Ile67fs), rrl (n.2270 G>T) and fbiC (Thr815fs), consistent with genotypic resistance to BDQ, LZD and Pa, respectively. During longitudinal follow-up, additional resistance-associated variants were detected predominantly within Rv0678, including Glu49fs, Ser220Asn, Leu365Phe, Tyr468* and Ala153Val, resulting in progressive diversification of the resistance genotype. An additional fbiC frameshift mutation (Glu814fs) was first detected in the third research isolate, collected in January 2025 during month 4 of the individualized treatment episode (TE2, M4), and persisted thereafter.

Longitudinal phenotypic susceptibility testing of the sequential research isolates demonstrated persistently elevated MICs to all three BPaL drugs throughout the study period. BDQ MICs were 1.0 μg/mL in isolates 1–3, collected during months 1–4 of the individualized pre-XDR-TB treatment episode (TE2, M1–M4), and increased to 2.0 μg/mL from isolate 4 (March 2025; TE2, M6) onwards, coinciding with increasing complexity of the Rv0678 mutational profile. In contrast, Pa MICs remained consistently above the highest concentration tested (>2.0 μg/mL), while linezolid MICs remained stable at 2.0 μg/mL across all isolates. Collectively, these findings demonstrate persistent multidrug resistance accompanied by progressive within-host genomic diversification, particularly within Rv0678, during the treatment course (Table 4).

Table 4.

Longitudinal phenotypic and genomic characteristics of seven sequential M. tuberculosis isolates.

Isolate	Collection date	Treatment episode / month	BDQ MIC (µg/mL)	Pa MIC (µg/mL)	LZD MIC (µg/mL)	Rv0678variants	fbiCvariants	rrlvariant
NM3605_UL19	Oct 2024	TE2, M1*	1.0	> 2.0	2.0	Ile67fs	Thr815fs	n.2270 G>T
ON8123_UL19	Dec 2024	TE2, M3*	1.0	> 2.0	2.0	Ile67fs; Glu49fs	Thr815fs	n.2270 G>T
NM1529_UL19	Jan 2025	TE2, M4*	1.0	> 2.0	2.0	Ile67fs; Glu49fs; Ser220Asn	Thr815fs; Glu814fs	n.2270 G>T
ON5717_UL19	Mar 2025	TE2, M6*	2.0	> 2.0	2.0	Ile67fs; Glu49fs; Ser220Asn; Leu365Phe	Thr815fs; Glu814fs	n.2270 G>T
ON7009_UL19	May 2025	TE2, M8*	2.0	> 2.0	2.0	Ile67fs; Glu49fs; Ser220Asn; Leu365Phe; Tyr468*	Thr815fs; Glu814fs	n.2270 G>T
NM7738_UL19	Jun 2025	TE2, M9*	2.0	> 2.0	2.0	Ile67fs; Glu49fs; Ser220Asn; Leu365Phe; Tyr468*; Ala153Val	Thr815fs; Glu814fs	n.2270 G>T
NM9142_UL19	Jul 2025	TE2, M10*	2.0	> 2.0	2.0	Ile67fs; Glu49fs; Ser220Asn; Leu365Phe; Tyr468*; Ala153Val	Thr815fs; Glu814fs	n.2270 G>T

TE2, second treatment episode, corresponding to the individualized pre-XDR-TB treatment initiated following microbiologically confirmed recurrence in October 2024; M, treatment month; BDQ, bedaquiline; Pa, pretomanid; LZD, linezolid; MIC, minimum inhibitory concentration; fs, frameshift. Treatment months are assigned using October 2024 as month 1 based on the documented month of initiation of the individualized treatment regimen; exact treatment initiation and isolate-collection dates were unavailable.

Outcome

Despite treatment with a guideline-based BPaL regimen and intensive supportive care, the patient deteriorated rapidly and demised from septic shock secondary to advanced drug-resistant tuberculosis. Routine phenotypic drug susceptibility testing became available only after her death and confirmed resistance to BDQ, Pa and LZD. Longitudinal phenotypic and genomic analyses of sequential research isolates further demonstrated persistent multidrug resistance with progressive diversification of Rv0678 during follow-up.

Beyond its clinical significance, this case provided a rare opportunity to integrate longitudinal clinical, phenotypic and genomic data from a single patient, enabling detailed characterisation of within-host resistance dynamics during treatment with a BPaL-based regimen.

Discussion

This case provides a rare longitudinal characterisation of fatal recurrent RR-TB by integrating detailed clinical information with serial phenotypic drug susceptibility testing and WGS of seven sequential M. tuberculosis isolates collected during treatment. While resistance to BDQ, Pa and LZD has increasingly been reported following implementation of BPaL-based regimens, reports describing the interaction between clinical progression, phenotypic resistance and within-host genomic evolution remain scarce. The novelty of this case lies in the integration of patient-related, microbiological and genomic data, providing a understanding of treatment failure that extends beyond routine clinical observation.

The patient's poor outcome was most likely the consequence of multiple interacting factors rather than a single event. These included recurrent drug-resistant disease, interruption of treatment following loss to follow-up, non-disclosure of pregnancy, delayed recognition of resistance to newer anti-tuberculosis drugs, advanced cavitary pulmonary disease at re-presentation, and progressive within-host evolution of a highly resistant M. tuberculosis strain. This multifactorial interaction highlights the complexity of managing DR-TB in routine programmatic settings, where biological, clinical and health-system factors collectively influence treatment outcomes [21], [22].

An important patient-related factor was prolonged interruption of treatment after the patient became pregnant. Treatment interruption is a well-recognised determinant of unfavourable tuberculosis outcomes because it allows continued bacterial replication, progressive pulmonary destruction and persistent transmission. In patients with DR-TB, inconsistent drug exposure may also create fluctuating antimicrobial selection pressures that favour persistence and diversification of resistant bacterial subpopulations [23], [24]. Although the temporal sequence of resistance acquisition cannot be established in this patient because the research isolates were collected after treatment had already commenced, interruption of therapy may have facilitated expansion of an already resistant bacterial population while allowing progressive accumulation of additional adaptive mutations. Had the patient remained engaged in care, uninterrupted clinical monitoring would likely have provided greater opportunity to detect persistent treatment failure, repeat microbiological investigations and consider alternative treatment strategies before she returned with extensive cavitary disease. While it cannot be concluded that uninterrupted treatment would necessarily have prevented the fatal outcome, continued engagement with specialised care would almost certainly have improved opportunities for earlier clinical intervention. Our observations are consistent with a recent case report from Zambia in which treatment interruption and delayed re-engagement with care were similarly associated with progressive amplification of resistance during BPaL-based therapy [20]. In both cases, prolonged interruption of treatment occurred alongside delayed drug susceptibility testing, illustrating how programmatic and patient-related factors may converge to compromise treatment outcomes.

Pregnancy further complicated the clinical course. The patient's decision not to disclose her pregnancy resulted in interruption of follow-up and delayed re-engagement with tuberculosis services until after delivery. Management of pregnant women receiving treatment for DR-TB requires coordinated multidisciplinary care involving TB clinicians, obstetricians and neonatal services to balance maternal treatment, fetal wellbeing and medication safety [25], [26]. Earlier disclosure of pregnancy may have facilitated adherence counselling, closer clinical review, enhanced maternal support and continued microbiological surveillance throughout pregnancy. Although it is impossible to determine retrospectively whether earlier disclosure would have altered the patient's ultimate outcome, the sequence of events highlights the importance of integrating maternal health and TB services to minimise interruptions in treatment among this vulnerable patient population.

Beyond patient-related factors, this case exposes an important limitation within current programmatic management of DR-TB. At the time of the patient's final admission, clinicians initiated BPaL regimen based on historical drug susceptibility results because routine phenotypic susceptibility testing for BDQ, Pa and LZD were not available. Resistance to all three drugs was confirmed only after the patient's death. This delay demonstrates a critical diagnostic gap in contemporary drug-resistant tuberculosis programmes. While rapid molecular assays are widely available for rifampicin, isoniazid and fluoroquinolone resistance, comparable rapid diagnostics for resistance to newer anti-tuberculosis drugs remain limited. Consequently, patients may continue receiving ineffective therapy despite appropriate adherence to national treatment guidelines. Similar diagnostic challenges have recently been described in Zambia, where delayed phenotypic DST results resulted in prolonged empirical treatment and delayed recognition of acquired BDQ and LZD resistance [20]. Together, these cases emphasise that improving laboratory turnaround times is essential for preserving the effectiveness of newer anti-tuberculosis agents under routine programmatic conditions. This case therefore reinforces the need for expanded access to rapid molecular and phenotypic susceptibility testing for BDQ, Pa and LZD to enable earlier identification of resistance and timely individualisation of therapy.

The longitudinal genomic findings provide further insight into the biological mechanisms underlying treatment failure. The first research isolate, collected in October 2024 during month 1 of the individualized pre-XDR-TB treatment episode (TE2, M1), already harboured resistance-associated variants in Rv0678, rrl and fbiC, indicating that resistance to all three BPaL drugs had already been established by the time research sampling commenced. Nevertheless, sequential isolates demonstrated continued accumulation of additional Rv0678 variants together with a two-fold increase in BDQ MIC. These observations are consistent with the established role of Rv0678 as the principal transcriptional regulator of the MmpS5-MmpL5 efflux system, where loss-of-function mutations increase drug efflux and reduce intracellular concentrations of BDQ. Rather than representing a single resistance event, the progressive accumulation of multiple Rv0678 variants suggests continued within-host adaptation under ongoing antimicrobial selection pressure after phenotypic BDQ resistance had already been established. This indicates that crossing the phenotypic resistance threshold did not represent an evolutionary endpoint, as the bacterial population continued to accumulate additional Rv0678 variants during treatment exposure, accompanied by a two-fold increase in BDQ MIC from 1.0 to 2.0 µg/mL. [17], [27].

Interestingly, the evolutionary trajectories differed between the three BPaL drugs. Whereas BDQ resistance was characterised by progressive diversification of Rv0678, the LZD-associated rrl mutation remained unchanged throughout follow-up and Pa resistance demonstrated early acquisition of an additional fbiC frameshift mutation followed by persistently elevated MICs. These findings suggest that resistance evolution is drug-specific rather than uniform across multidrug regimens. Different resistance mechanisms are likely subjected to distinct selective pressures and biological constraints, resulting in heterogeneous evolutionary trajectories even within the same infecting strain [28], [29]. Such observations reinforce the importance of considering each resistance mechanism independently when interpreting longitudinal genomic data.

An equally important observation concerns the relationship between genomic evolution and phenotypic resistance. Although additional resistance-associated mutations accumulated during follow-up, corresponding changes in phenotypic susceptibility were relatively modest. BDQ MICs increased from 1.0 µg/mL to 2.0 µg/mL as additional Rv0678 mutations accumulated, whereas Pa and LZD MICs remained consistently elevated despite comparatively stable resistance genotypes. These findings demonstrate that increasing genomic complexity does not necessarily translate into proportionate increases in phenotypic resistance. Instead, resistance evolution appears to involve continued accumulation of adaptive mutations that may reinforce existing resistance mechanisms, improve bacterial fitness or compensate for fitness costs associated with earlier mutations without producing large incremental changes in MIC [28]. This observation further supports integrated interpretation of phenotypic and genomic data rather than reliance on a single approach.

Although the retrospective WGS analyses did not influence the patient's clinical management, they provided important biological insights that would not have been obtained through routine diagnostics alone. The integration of serial MIC measurements with longitudinal genomic surveillance demonstrated persistent multidrug resistance accompanied by progressive within-host diversification of Rv0678, illustrating that resistance evolution remains a dynamic process even after clinically significant resistance has become established. These findings support the incorporation of genomic surveillance into prospective monitoring of patients receiving newer anti-tuberculosis agents, particularly in high-burden settings where resistance mechanisms continue to emerge and conventional susceptibility testing remains limited.

Although acquired resistance to drugs in the BPaL regimen under routine programmatic conditions has been described, most have focused primarily on clinical progression and final resistance profiles. For example, the recent case from Zambia documented treatment interruption, delayed drug susceptibility testing and subsequent emergence of dual BDQ and LZD resistance during BPaL therapy [20]. This case extends these observations by integrating longitudinal clinical information with serial MIC measurements and WGS of seven sequential isolates, enabling detailed characterisation of within-host genomic diversification over time. Rather than documenting resistance at a single endpoint, this study illustrates the dynamic relationship between patient-related factors, health-system limitations, phenotypic resistance and progressive bacterial adaptation throughout treatment.

The strength of this report lies in the integration of clinical observations with sequential phenotypic and genomic characterisation from a single patient. Rather than describing resistance at a single time point, the availability of seven sequential research isolates enabled detailed examination of within-host bacterial adaptation over time. Nevertheless, several limitations should be acknowledged. The research isolates were collected during treatment rather than from the initial diagnostic episode, preventing determination of the precise timing of resistance acquisition in this patient. Although this report describes a single patient and cannot be generalised to all patients receiving BPaL therapy, it provides a biologically plausible framework for understanding how treatment interruption may influence resistance dynamics during therapy with newer anti-tuberculosis agents. The integration of clinical, phenotypic and genomic data suggests that prolonged interruption of treatment may permit ongoing bacterial replication and continued within-host diversification of resistant populations while simultaneously delaying recognition of treatment failure. These observations have broader implications for DR-TB programmes, emphasising that the long-term effectiveness of BPaL regimens depends not only on the potency of the drugs themselves, but also on sustained patient retention, uninterrupted treatment, timely resistance testing and integrated genomic surveillance. An additional limitation was the incomplete availability of precise dates for microbiological investigations during the earlier treatment episode. Although persistent bacteriological positivity was documented following the initial RR-TB diagnosis, followed by smear positivity (2 +) and newly acquired fluoroquinolone resistance in August 2023, the exact dates of the preceding positive microbiological results, specimen collection and reporting of DST results were unavailable. Consequently, the interval between the initial indication of poor treatment response, subsequent sampling and availability of DST results during this earlier episode could not be reliably quantified.

In conclusion, this case demonstrates that treatment failure during BPaL therapy cannot be explained solely by bacterial resistance or patient behaviour in isolation. Rather, it reflects the interaction of host-related factors, health-system limitations and progressive bacterial adaptation within the host. By integrating clinical information, treatment history, phenotypic susceptibility and longitudinal genomic evolution, this report illustrates how interruption of care, delayed recognition of resistance and continued within-host diversification of M. tuberculosis may converge to produce poor clinical outcomes. These observations provide a patient-level perspective on resistance evolution and underscore the importance of integrated phenotypic and genomic surveillance, uninterrupted patient retention and timely drug susceptibility testing for newer anti-tuberculosis agents in optimising management of drug-resistant tuberculosis

CRediT authorship contribution statement

Kabelo Gabriel Kaapu: Writing – review & editing, Writing – original draft, Methodology, Investigation, Formal analysis, Conceptualization. Ramathetje Virginia Ntsoane: Writing – review & editing, Resources, Methodology, Investigation. Casious Raphiri: Writing – review & editing, Resources, Investigation. Molebogeng Ruth Lekalakala-Mokaba: Writing – review & editing, Resources, Project administration, Methodology. Ivy Rukasha: Writing – review & editing, Supervision, Methodology, Investigation, Conceptualization.

Informed Consent Statement

Informed consent was waived by the Limpopo Department of Health as this study was based on a retrospective review of anonymized clinical and laboratory data. No direct contact was made with the patient for the purpose of this case report.

Institutional Review Board Statement

Ethical approval was granted by the Turfloop Research Ethics Committee (TREC/108/2024 PG) and the National Health Laboratory Service Academic Affairs (PR2456259, approval date 10 January 2024). Limpopo department of health authorised the study (LP_2025–03–011).

Ethics declaration

The authors have NOT obtained informed consent from participants or their legal representatives. Patient investigation was done after the patient had demised.

This study was performed in compliance with relevant laws, regulatory frameworks and guidelines where the research took place. This study was approved by the Turfloop Research Ethics Commitee. (Approval No. TREC/108/2024 PG)

This research follows the CARE guidelines and the CARE checklist.

Funding

The authors declare that no financial support was received for the research.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability

The datasets presented in this study can be found in online repositories. The raw fastq files were submitted to the National Center for Biotechnology Information (NCBI) under the BioProject ID PRJNA1416012.