Leveraging rpoB amplicon signals from Xpert MTB/RIF ultra as a diagnostic opportunity for identifying Non-tuberculous mycobacteria in TB-negative specimens

Abstract

Non-tuberculous mycobacteria (NTM) present a diagnostic challenge due to clinical overlap with tuberculosis (TB). The Xpert MTB/RIF Ultra assay occasionally detects rpoB probe amplification in specimens negative for the Mycobacterium tuberculosis complex (MTBC). We evaluated whether rpoB amplification could indicate NTM infection and support diagnostic decision-making. Specimens showing rpoB probe amplification but negative for MTBC by Xpert Ultra were selected from TB Screening and Treatment Centers (TBSTC) of icddr, b between January and August 2024. Samples underwent further testing using PCR targeting rpoB and IS6110, culture, and targeted next-generation sequencing (tNGS) to detect and identify NTM species. Among 50 specimens with rpoB amplification, rpoB PCR positivity was confirmed in all cases. Atypical mycobacterial growth was observed in 43 (86%) isolates. tNGS identified 34 isolates as NTM, with Mycobacterium gordonae (20.5%) and Mycobacterium abscessus subsp. abscessus (17.6%) being the most prevalent; 3 were MTBC, and the 6 were not detected. Receiver operating characteristic (ROC) analysis of rpoB2 threshold cycle (Ct) values demonstrated moderate diagnostic performance (AUC 0.807) with a Ct cutoff of 32.7 yielding 82.4% sensitivity, 77.8% specificity, and a positive predictive value of 93.3%. At a cutoff value of 32.7, the sensitivity and specificity for pulmonary samples were 90% and 100%, respectively, whereas for extrapulmonary samples they were 76% and 85%, respectively. These findings indicate that rpoB2 Ct values from the Xpert Ultra assay could provide additional diagnostic utility for identifying NTMs in MTBC-negative samples without additional costs.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1038/s41598-026-57683-z.

Introduction

Tuberculosis (TB) is a major global health threat which is caused by the Mycobacterium tuberculosis complex (MTBC)¹. However, non-tuberculous mycobacteria (NTM) are increasingly being recognized as an emerging public health concern. There are several studies which reported a rising incidence of NTM infections worldwide². NTM pulmonary diseases (NTM-PD) mimic the TB and other pulmonary conditions with non-specific symptoms and radiographic features^3,4. Both MTBC and NTM can also infect extrapulmonary sites. However, the clinical manifestations of these infections vary based on species, infection sites, and host immune status. This clinical overlap can lead to misdiagnosis in high TB burden countries, where TB is often presumed.

Critically, NTM are intrinsically resistant to many antibiotics, including first-line anti-TB drugs. Misdiagnosing NTM infections as TB—especially drug-resistant TB— can delay appropriate treatment and increase morbidity. Misdiagnosis may also lead to unnecessary patient isolation, psychological distress, stigma, and inappropriate public health responses, including outbreak investigations. Management of both pulmonary and extrapulmonary NTM infections typically requires prolonged, complex antibiotic regimens, and even treatment; mortality remains high. A five-year cohort study of over 9,000 NTM-PD patients reported a 27% case fatality rate (95% CI: 21.3–37.8)^5,6. In Bangladesh, NTM prevalence is not well established due to the limited number of studies or resource limitations. However, a study conducted in a tertiary care hospital found a total of 7.7% of NTM prevalence among suspected tuberculosis patient with non-healing surgical site infections, and the most prevalent NTM species were Mycobacterium abscessus and M. fortuitum⁷.

GeneXpert MTB/RIF Ultra assay (Xpert Ultra) has transformed TB diagnosis by enabling rapid detection of MTBC and rifampicin resistance⁸. The assay identifies MTBC by targeting the multicopy insertion sequences IS6110 and IS1081 and detects RIF resistance through melt curve analysis of four probes (rpoB1-rpoB4) targeting the rpoB gene (Fig. 1).

Fig. 1.

Schematic representation of the four molecular beacon probes (rpoB1-rpoB4) of rpoB gene employed in the Xpert ultra-assay. The probes contain overlapping sequences and produce different melting temperatures to detect RIF-R.

“MTB Detected” occurs when IS6110/IS1081 and at least one rpoB gene target is amplified. If at least two rpoB targets are also detected, the assay provides a semi-quantitative estimate of bacterial load. Conversely, if only IS6110/IS1081 is detected and rpoB amplification does not occur, the assay reports “MTBC Trace Detected”. “MTB Not Detected” is reported by the Xpert ultra if IS6110/IS1081 multi-copy gene is not amplified, regardless of rpoB probe amplification. The Xpert Ultra assay primarily uses rpoB probes to detect rifampicin resistance within MTBC. Amplification of rpoB probes in the absence of IS6110/IS1081 signals does not meet the assay criteria for MTBC detection. One or more rpoB probe signals are often observed in “MTB Not Detected” samples that may poses a diagnostic dilemma, where MTBC and NTM share partial sequence similarity in rpoB gene. Hence, symptomatic patients with rpoB amplification, but no MTBC-specific signal may be overlooked, despite the possibility of an NTM infection. In settings with limited access to culture or sequencing, these cases may receive empiric TB treatment or go undiagnosed. While some studies have investigated the potential for Xpert Ultra to misclassify NTM as MTBC in high bacterial load samples^9,10, few have explored the diagnostic significance of rpoB amplification in Xpert Ultra-negative specimens. A recent study¹¹ reported a preliminary NTM prevalence of 1.7% in Xpert Ultra-negative cases, highlighting the potential for under-recognition of NTM infections.

Given the low prevalence of NTM, routine testing of all Xpert Ultra negative specimens is not feasible in resource-limited settings. However, reports of Xpert Ultra misclassifying high-burden NTM infections as MTBC due to rpoB probe amplification^9,10 suggest that rpoB signals in the absence of IS6110/IS1081 detection may indicate the presence of NTM. Despite the potential clinical significance, comprehensive studies examining IS1081/IS6110-negative, rpoB-positive (Xpert Ultra-negative) cases remain limited¹².

To address this gap, we investigated specimens that were Xpert Ultra negative for MTBC but showed amplification of the rpoB region. This study aims to improve the identification of samples warranting NTM testing by leveraging data already generated by the standard-of-care test, the Xpert Ultra, in Bangladesh.

Materials and methods

Study design and sample collection

This prospective study analyzed diagnostic specimens from patients referred to the TB Screening and Treatment Center (TBSTC) of the International Centre for Diarrhoeal Disease Research, Bangladesh (icddr, b) between January and August 2024. Patients were referred by private clinicians for confirmatory testing based on clinical, radiological, or symptomatic evidence suggestive of pulmonary (PTB) or extrapulmonary TB (EPTB). Pulmonary samples consisted of sputum, while extrapulmonary samples included tissue, urine, bronchoalveolar lavage (BAL) fluid, pus, synovial fluid, and stool. Upon receipt, all specimens were tested using the Xpert Ultra assay at the Mycobacteriology Laboratory. All methods in the study were performed in line with the relevant guidelines and principles of the Declaration of Helsinki. Written informed consent was obtained from all participants and/or their legal guardian(s) prior to sample collection.

Sample selection criteria

Specimens with a negative Xpert Ultra result (i.e., no amplification of IS6110/IS1081) but with amplification of at least one rpoB probe (rpoB1, rpoB2, rpoB3, or rpoB4) were considered for inclusion. During the study period, a total of 65 samples were identified having such criteria. Of these, 15 samples were excluded due to a lack of sufficient sample volume (< 2 mL) for further tests. The remaining 50 specimens were included in the downstream analysis.

Laboratory procedures

Conventional polymerase chain reaction (PCR) targeting the rpoB and IS6110 genes was performed for all 50 samples, and specimens were simultaneously cultured on Lowenstein–Jensen (L-J) media. Culture-positive isolates underwent targeted next-generation sequencing (tNGS) to identify NTM species. The diagnostic workflow of the different test methods is shown in Fig. 2. Final classification of specimens as MTBC or NTM was based on tNGS results, which provided species-level identification through targeted sequencing of the mycobacterial hsp65 gene region. In cases of discordant results between Xpert Ultra, PCR, and culture, tNGS was considered the definitive reference standard due to its species-level resolution capability.

Fig. 2.

Study workflow of specimen selection and diagnostic testing. IS6110-negative and rpoB2-positive specimens identified by Xpert Ultra between January and August 2024 were included in the study. Available specimens were subjected to conventional PCR and culture. tNGS was subsequently performed on culture-positive isolates to determine the presence of mycobacterial species.

Ethical considerations

The protocol was approved by the Institutional Review Board (IRB) of International Centre for Diarrhoeal Disease Research, Bangladesh (icddr, b), which includes the Research Review Committee (RRC) and Ethical Review Committee (ERC).

Xpert MTB/RIF ultra assay

The Xpert Ultra assay was performed on sputum samples according to the manufacturer’s instructions¹³. Standard sample processing protocols were followed for the extrapulmonary samples considering the specific specimen type and volume¹⁴. Stool specimens were prepared using a previously established standardized method¹⁵. The processed samples were mixed with SR in a 1:2 ratio, incubated with intermittent mixing, and 2 mL of the processed sample was loaded into the Ultra cartridge for automated analysis.

Specimen processing for culture

Residual unprocessed specimens were decontaminated using the standard N-acetyl-L-cysteine–sodium hydroxide (NALC-NaOH) method¹⁶. Following decontamination, the resulting pellet was resuspended in 1.0 mL of phosphate-buffered saline (PBS). Two loopfuls of pellets were inoculated onto L-J slants for visible growth.

Conventional PCR targeting IS6110 and rpoB genes

DNA extraction was performed using the heat sonication method as previously described¹⁷. Conventional rpoB PCR was conducted to validate amplification of the rpoB probe detected in the Xpert Ultra assay¹⁸. IS6110 PCR was performed to confirm the presence or absence of MTBC¹⁸.

Targeted next generation sequencing (tNGS)

All culture-positive isolates with atypical colony morphology underwent tNGS for species level identification. Atypical morphology was defined as colonies that were smooth, pigmented, rapidly growing, or lacking the characteristic rough, buff appearance of MTBC.

DNA extraction was performed using the standard protocol described previously¹⁹. PCR amplification of target regions was carried out with the Genoscreen Deeplex Myc-TB kit. Library preparation and sequencing were performed using the Illumina MiniSeq platform following the manufacturer’s instructions¹⁹. Data analysis was performed using GenoScreen’s cloud-based Deeplex Myc-TB tool (Version 3.0), which identified NTM species based on the hsp65 gene region.

Statistical analysis

All statistical analyses were performed using R software (version 4.5.2). Continuous variables were summarized as median with interquartile range (IQR) and categorical variables as counts and percentages. Differences in rpoB2 Ct values among different NTM groups identified by Kruskal-Wallis test. Differences in rpoB2 Ct values between NTM and MTB/ND groups identified by tNGS were assessed using the Wilcoxon rank-sum test. Outliers or influential observations were identified using Cook’s distance. Diagnostic performance of rpoB2 was evaluated using Receiver Operating Characteristic (ROC) curve analysis, and the area under the curve (AUC) with 95% confidence intervals (CI) was calculated using the DeLong method. The optimal Ct cutoff was determined using the Youden index. Internal validation of ROC estimates was performed using bootstrap resampling (2000 stratified replicates). The diagnostic performance was compared before and after excluding influential observation. Comparisons between pulmonary and extrapulmonary specimens were performed using the Chi-square or Fisher’s exact test for categorical variables and the Wilcoxon rank-sum test for Ct values. Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) with 95% confidence intervals were calculated for pulmonary and extrapulmonary samples using the predefined Ct cutoff. A p-value < 0.05 was considered statistically significant.

Results

Socio-demographic, clinical characteristics, and specimen types

A total of 50 samples that were negative for both IS6110/IS1081 and positive for at least one rpoB probe in the Xpert Ultra assay were included in the analysis. Patient demographic characteristics are summarized in Table 1. The most represented age group was individuals > 60 years, comprising 40% (20/50) of the total participants. In contrast, the ≤ 20 years age group accounted for only 4% (2/50). Overall, the median age of all participants was 61 years (IQR: 45.5–65.0) and males made up 56% (28/50) of the study population. Sputum samples represented 50% (25/50) of all samples. Among extrapulmonary specimens, tissue was the most common (13/50, 26%), followed by urine (6/50, 12%) and BAL fluid (3/50, 6%). Other extrapulmonary specimens, such as pus, synovial fluid, and stool, each accounted for 2% (1/50).

Table 1.

Demographic and clinical characteristics of study samples.

Characteristics	Category	No. of Patients (N = 50)
Age (years)	< 20	2 (4%)
	21–30	6 (12%)
	31–40	6 (12%)
	41–50	8 (16%)
	51–60	8 (16%)
	> 60	20 (40%)
Sex	Female	22 (44%)
	Male	28 (56%)
Specimen types	Pulmonary
	Sputum	25 (50%)
	Extrapulmonary
	BAL fluid	3 (6%)
	Tissue	13 (26%)
	Urine	6 (12%)
	Pus	1 (2%)
	Synovial fluid	1 (2%)
	Stool	1 (2%)

Xpert ultra results

All 50 specimens included in this study were negative for IS6110/IS6110 (Ct > 40) but showed amplification of at least one rpoB probe (Ct < 40) in the Xpert Ultra assay. Probe-specific detection rates were as follows: rpoB1 in 14% (7/50), rpoB2 in 100% (50/50), rpoB3 in 4% (2/50), and rpoB4 in 52% (26/50). Nearly half of the samples 48% (24/50) showed rpoB2 amplification alone, while the remaining 52% (26/50) showed combinations of rpoB2 with one or more additional rpoB probes. Specifically, the combinations included:

rpoB2 + rpoB4, rpoB1 + rpoB2+rpoB4, and rpoB1 + rpoB2+rpoB3 + rpoB4. The distribution of rpoB probe combinations is illustrated in Fig. 3A, while representative Ct values for selected probes are shown in Fig. 3B.

Fig. 3.

(A) Distribution of rpoB probes (rpoB1, rpoB2, rpoB3, and rpoB4) detected in Xpert Ultra assay positive samples. All samples were IS6110 negative and positive for at least one rpoB probe. (B) Xpert Ultra test results showing an example of Xpert Ultra negative, Ct values of rpoB2 and rpoB4 probes without IS6110/IS1081 amplification.

Comparison of diagnostic performance: Xpert Ultra vs. PCR, Culture, and tNGS

The diagnostic performance of the Xpert Ultra assay was evaluated in 50 specimens that were negative for IS6110 but positive for at least one rpoB probe. Conventional PCR, culture, and tNGS results of these samples were summarized in Table 2. All 50 samples were positive for rpoB gene in PCR, demonstrating 100% concordance with the rpoB signals detected by Xpert Ultra. In contrast, 88% (44/50) were negative for IS6110 gene in PCR, consistent with Xpert Ultra “MTB Not Detected” result and the remaining 12%, (6/50) were positive.

Table 2.

PCR, culture, and tNGS results for samples collected from presumptive TB patients.

Test	Pulmonary samples		Extrapulmonary samples		Total positive (n%)	Total negative (n%)
	Positive	Negative	Positive	Negative
rpoB PCR (n = 50)	25	0	25	0	50 (100%)	0 (0%)
IS6110 PCR (n = 50)	4	21	2	23	6 (12%)	44 (88%)
Culture (n = 50)	23	2	20	5	43 (86%)	7 (14%)
tNGS (n = 43)	22	1	15	5	34 (79%) NTM, 3 (7%) MTBC]	6 (14%)

86% (43/50) of samples were culture-positive, and culture positivity did not differ significantly between pulmonary and extrapulmonary samples (p = 0.4174). All culture-positive isolates showed atypical colony morphology, suggesting that they were NTMs or other bacterial species. Subsequent tNGS identified 79% (34/43) of culture-positive isolates as NTMs, 7% (3/43) as MTBC, and 14% (6/43) as not detected (ND) due to insufficient mycobacterial reads or inadequate sequencing coverage for species-level identification by Deeplex^® Myc-TB.

Overall 68% (34/50) of the 50 Xpert Ultra-negative but rpoB-positive samples were confirmed as NTM by tNGS. These findings strengthen the diagnostic value of incorporating PCR, culture, and sequencing together to accurately identify NTMs in clinical samples where Xpert Ultra results are negative for MTBC but show rpoB amplification. The remaining 7 (14%, 7/50) samples that were culture-negative were excluded from further testing. All test results of the corresponding samples are represented in Supplementary Table 1.

Distribution of NTM species detected by tNGS

A total of 34 NTM isolates were identified, representing different species or subspecies levels (Table 3). M. gordonae was the most prevalent species, found in 20.5% (7/34) of cases, mainly from the pulmonary samples (85.7%, 6/7) and one from an extrapulmonary source. M. abscessus subsp. abscessus was the second most frequent, making up 17.6% (6/34) of the total isolates. Unlike M. gordonae, this species was more often found in extrapulmonary samples (4/6) than pulmonary (2/6). M. fortuitum subsp. acetamidolyticum or M. fortuitum was another frequently detected species (17.6%, 6/34), predominantly appearing in extrapulmonary specimens (4/6). Overall, pulmonary specimens accounted for 61.7% (21/34) of NTM isolates, while 38.2% (13/34) came from extrapulmonary sources (Supplementary Fig. 1 and Supplementary Table 2). Among 13 NTMs detected in extrapulmonary specimens, 53.8% (7/13) were from tissue, and 15.4% (2/13) from urine and BAL samples. Tissue specimens comprised of three M. abscessus subsp. abscessus, three M. fortuitum subsp. acetamidolyticum or M. fortuitum and one M. chubuense or M. chlorophenolicum. Urine isolates included M. gordonae and M. asiaticum, while BAL comprised M. brisbanense and M. fortuitum subsp. acetamidolyticum or fortuitum.

Table 3.

Types and frequency of NTM species identified by tNGS in pulmonary and extrapulmonary specimens.

Species	Count (n = 34; %)	Culture Positive (n)	IS6110 Positive (n)	rpoB Positive (n)
M. gordonae	7 (20.5)	7	3	7
M. abscessus subsp. abscessus	6 (17.6)	6	0	6
M. asiaticum	4 (11.7)	4	1	4
M. brisbanense	2 (5.8)	2	0	2
M. arupense	1 (2.9)	1	1	1
M. insubricum	1 (2.9)	1	0	1
M. vanbaalenii	1 (2.9)	1	0	1
M. fortuitum subsp. acetamidolyticum or fortuitum	6 (17.6)	6	1	6
M. gordonae or asiaticum	2 (5.8)	2	0	2
M. abscessus subsp. abscessus or M. houstonense or M. fortuitum subsp. acetamidolyticum or fortuitum	1(2.9)	1	0	1
M. chubuense or M. chlorophenolicum	1(2.9)	1	0	1
M. nonchromogenicum or M. arupense	1(2.9)	1	0	1
M. agri	1(12.9)	1	0	1

Analysis of Xpert Ultra Ct values for differentiation of MTBC and NTM

Of 34 NTMs identified by tNGS, 41.2% (14/34) amplified rpoB2 alone, one amplified all four probes, six isolates demonstrated amplification of rpoB1, rpoB2 and rpoB4, and the remaining 13 showed amplification of rpoB2 with rpoB4. The median Ct for the rpoB1 probe was 34.5 (IQR: 33.2–37.2), while rpoB2 showed the lowest variability among positive signals with a median of 27.2 (IQR: 19.8–32.3). The rpoB4 probe had a median Ct of 33.2 (IQR: 26.7–36.3) across 20 NTM species. The lowest median Ct values of rpoB2 were observed for M. gordonae (18.3; IQR: 17.8–20.3) and M. gordonae or M. asiaticum (18.4; IQR: 17.7–19.1), indicating strong amplification for these species. M. brisbanense showed the highest median Ct (32.4; IQR: 31.7–33.0), followed by M. abscessus subsp. abscessus (30.8; IQR: 28.3–34.7). Intermediate Cts were observed for M. fortuitum subsp. acetamidolyticum or fortuitum (28.7; IQR: 26.4–31.7) and M. asiaticum (19.6; IQR: 18.1–22.1). Since rpoB2 was amplified in most of the NTM species, we compared the median Ct using the Boxplot and Kruskal-Walli’s test (Fig. 4A). There was a statistically significant difference in the median Ct value for rpoB2 among the NTM species (X² = 21.632, p = 0.04185). The rpoB2 Ct values for single M. arupense, M. insubricum, M. vanbaalenii, M. abscessus subsp. abscessus or M. houstonense or M. fortuitum subsp. acetamidolyticum or fortuitum, M. chubuense or M. chlorophenolicum, M. nonchromogenicum or M. arupense, and M. agri isolates were 20.7, 31.5, 32.2, 39.8, 32.6, 24.9 and 18.1, respectively.

Fig. 4.

A) Box plot for Ct values of rpoB2 for individual NTM species identified by tNGS. The lowest median rpoB2 Ct (18.3) values were found in M. gordonae and related subspecies, while the highest median Ct (32.4) was found in M. brisbanense. B) Evaluation of rpoB2 Ct values among NTM and MTB/ND groups based on tNGS results before excluding influential observations. C) Evaluation of rpoB2 Ct values among NTM and MTB/ND groups based on tNGS results after excluding influential observations. Median Ct value can potentially differentiate NTM from MTB/ND in both cases.

Since all 43 culture-positive isolates showed amplification of the rpoB2 probe, its potential to differentiate NTM from the combined MTB/ND group was evaluated. A Wilcoxon rank-sum test demonstrated a statistically significant difference (p < 0.05) in Ct values between these groups both before and after excluding two influential observations suggesting discriminatory utility of the rpoB2 Ct values. The distinctive distribution of rpoB2 Ct values between NTM (Ct Values ≤ 32) and MTB/ND (> 32) groups illustrated in Fig. 4B and Fig. 4C.

To further evaluate the diagnostic performance of rpoB2 Ct values in distinguishing NTM from MTB/ND, ROC curve analysis was performed, yielding an AUC of 0.807 (95% CI 0.63–0.98) (Fig. 5A), indicating moderate discriminatory ability. A Ct cutoff 32.7 distinguished NTM from MTB/ND samples with sensitivity of 82.4% (95% CI: 65.47–93.24%), specificity of 77.8% (95% CI: 39.99–97.19%), a PPV of 93.3% (95% CI: 77.93–99.18%), and an NPV of 53.9% (95% CI: 25.13–80.78%). After excluding two influential observations, diagnostic performance improved, with AUC increasing to 0.905 (95% CI 0.805–1.00) (Fig. 5B). The optimal Ct threshold remained stable at 32.7, with sensitivity 84.8% (95% CI: 68.1–94.89%) and specificity 87.5% (95% CI: 47.35–99.68%), indicating the robustness of the predictive model.

Fig. 5.

A) Area Under the curve (AUC) for Ct values of rpoB2 probe distinguishing between the NTM and MTB/ND groups before excluding influential observations. B) Area Under the curve (AUC) for Ct values of rpoB2 probe distinguishing between the NTM and MTB/ND groups after excluding influential observations.

A significant difference was observed in the median Ct values between pulmonary and extrapulmonary samples (p = 0.0080). The median Ct value for pulmonary samples was 20.6 (IQR: 18.3–29.1), whereas extrapulmonary samples had a higher median Ct value of 32.6 (IQR: 29.2–33.6). Using the estimated Ct cutoff of 32.7, the sensitivity and specificity for pulmonary samples were 90% and 100%, respectively, with a PPV of 100% and an NPV of 33%. For extrapulmonary samples, the sensitivity and specificity were 76% and 85%, respectively, with a PPV of 90% and an NPV of 66%.

Discussion

NTM often mimics the clinical and radiological features of tuberculosis (TB), interfering with accurate diagnosis. Given that presumptive TB patients referred to our TB screening centers for evaluation by Xpert Ultra, we hypothesized that MTBC not detected samples with rpoB probe amplification might contain NTMs. The findings from this study strongly support that hypothesis and offer a new approach for identification of NTM.

All identified NTM isolates showed at least one rpoB probe amplification in the Xpert Ultra assay, with rpoB2 amplification present in 100% of cases, either alone or in combination with other probes. Amplification of rpoB2 probe alone was found in 41.2% (14/34) of NTM isolates. Several other prior studies found similar findings. Ayinalem Alemu et al. reported rpoB2 amplification in 88.5% of NTM isolates in Ethiopia²⁰, and a South African study observed a similar pattern in 83.3% of NTM-positive cases¹⁰. This consistent rpoB probe amplification pattern indicates sequence homology between the rpoB regions of NTM and MTBC. Interestingly, in our study 58.8% (20/34) of NTM isolates showed amplification of two or more rpoB probes. This percentage is higher than the proportions reported in South Africa (33.3%) and Ethiopia (16.1%)^10,20. In our cohort, the most frequent combination was rpoB2 + rpoB4 (38.2%), followed by rpoB1 + rpoB2 + rpoB4 (17.6%) and the full set of rpoB1, rpoB2, rpoB3, and rpoB4 (2.9%), which may reflect geographic variability in circulating NTM species.

Among the 50 Xpert Ultra-negative but rpoB-positive samples, 68% (34/50) were confirmed as NTM by tNGS. This is a notably higher prevalence than the previous studies where they reported 1.7% NTMs in Xpert Ultra-negative samples¹¹. This disparity indicates the diagnostic value of further evaluation of rpoB-positive cases in Xpert ultra, rather than all Xpert Ultra-negative specimens, which would be impractical in resource-limited settings. These evaluations were performed using conventional PCR, culture, and tNGS, as outlined in the diagnostic workflow presented in Fig. 2.

The rpoB2 Ct value derived from the Xpert Ultra assay showed potential utility in differentiating NTM from MTB/ND samples. ROC analysis demonstrated a moderate discriminatory ability (AUC = 0.807). During the analysis, two influential observations were identified with Ct values of 39.8 and 18.6 corresponding to NTM and ND results, respectively, which deviated from the overall distribution of Ct values between the NTM and MTB/ND groups. Such discrepancies may reflect biological variability, differences in mycobacterial load, or technical factors affecting amplification efficiency and sequencing read depth. For example, a very high Ct value in an NTM sample may indicate a low mycobacterial DNA concentration near the detection limit, whereas a low Ct value in an ND sample may result from insufficient sequencing coverage or limited mycobacterial reads during tNGS analysis. Sensitivity analysis performed after excluding these observations showed a slight improvement in diagnostic performance (AUC = 0.9), while the optimal Ct cutoff remained stable at 32.7. This stability indicates that the predictive value of the rpoB2 Ct was robust and not driven by extreme observations, supporting its potential use as an additional indicator to facilitate early differentiation of NTM in samples where Xpert Ultra reports MTB not detected but rpoB probe signals are present.

We also observed a significant difference in Ct values between pulmonary and extrapulmonary specimens, with extrapulmonary samples showing higher Ct values overall. This difference may reflect lower mycobacterial DNA load or differences in specimen characteristics commonly observed in extrapulmonary samples. The diagnostic performance of the Ct cutoff varied between pulmonary and extrapulmonary specimens, with higher sensitivity observed in pulmonary samples. This suggests that the proposed Ct threshold may perform more reliably in pulmonary specimens.

The proposed rpoB2 Ct cutoff of 32.7 demonstrated a high PPV, suggesting potential utility as a rule-in indicator for NTM among Xpert Ultra MTBC-negative but rpoB-positive samples. However, the relatively modest NPV indicates that Ct values above the threshold cannot reliably exclude NTM infection. Therefore, samples near or above the threshold should undergo additional confirmatory testing, including culture, PCR, or tNGS where available.

Species of identification by tNGS revealed diverse NTM populations, with M. gordonae, M. abscessus subsp. abscessus, and M. asiaticum as the most common isolates. Species distribution varied by specimen type: M. gordonae and M. asiaticum were primarily pulmonary, whereas M. abscessus and M. fortuitum subsp. acetamidolyticum were more frequently found in extrapulmonary samples. The predominance of Mycobacterium gordonae and the absence of Mycobacterium avium complex (MAC), a major cause of NTM lung disease globally²¹ may reflect regional differences in NTM epidemiology or limitations in culture conditions. Currently, data on NTM prevalence in Bangladesh remain limited²². Because mycobacterial burden and species distribution may vary geographically, the proposed Ct threshold may not be directly generalizable to other settings. Therefore, local verification and recalibration of the Ct cutoff may be necessary before implementation in different epidemiological contexts.

Probe amplification patterns vary in different NTM species. For example, six of seven M. gordonae isolates amplified by both rpoB2 and rpoB4. Similar pattern was observed in studies from South Africa and Ethiopia^10,20. Similarly, three of six M. abscessus subsp. abscessus isolates showed amplification of rpoB1 and rpoB2, while all four M. asiaticum isolates amplified in both rpoB2 and rpoB4. Five of six M. fortuitum subsp. acetamidolyticum isolates showed rpoB2-only amplification. These consistent patterns suggest that certain rpoB probe combinations may help infer NTM species from Xpert Ultra data—a potentially valuable tool where sequencing is unavailable.

One of the unexpected findings in our study was the detection of conventional IS6110 PCR positivity in six tNGS-confirmed NTM isolates. Possible explanations include cross-reactivity, genomic similarity, limitations of different tests, laboratory contamination, or mixed infections²³. In tNGS testing, hsp65 gene region is targeted to differentiate between MTBC and NTM as well as NTM species identification. Regarding these unusual cases, there might be mixed infections with MTBC and NTM with predominating burden of NTMs that were identified by tNGS, whereas IS6110 PCR only amplified the region targeting the MTBC. Notably, repeat testing confirmed the initial results, indicating the need for further evaluation of test specificity and the genetic heterogeneity of NTMs in this region^24,25. There were three MTBC-confirmed samples which were negative IS6110 PCR. It is evident from previous studies that the presence of MTBC strains harboring very few copy numbers or even fully absence of IS6110 element in many regions of Asia and certain parts of the world^26,27. This might be the possible reason for MTBC confirmed by tNGS based on hsp65 gene region in our study that was IS6110 PCR negative. Of the six IS6110-positive NTM isolates, four were pulmonary, and two were extrapulmonary. Among the three IS6110-negative MTBC isolates, one was pulmonary, and two were extrapulmonary, suggesting no clear clustering by specimen type.

This study had several limitations. First, the relatively small sample size, particularly the limited number of MTB/ND control samples, may affect the precision of the estimated Ct cutoff and diagnostic accuracy measures. However, post hoc power analysis (power = 0.88) suggested adequate statistical power for detecting discriminatory performance, and 95% confidence intervals were provided to contextualize the precision of the estimates. Nevertheless, these findings should be considered exploratory and require validation in larger, prospectively powered cohorts. In addition, the lack of detailed clinical data limited our ability to correlate NTM species with disease severity or clinical relevance. Second, this study is conducted in a single-centered setting; however, people with presumptive TB are usually referred for TB testing to this reference laboratory where more than 700 Xpert Ultra testing is performed monthly. Third, reliance on Lowenstein–Jensen (L-J) medium instead of MGIT media, due to budget constraints, may have restricted the diversity of NTM species isolated. Alternative media such as Middlebrook 7H10/7H11 and the combined use of solid and liquid culture could improve recovery of fastidious mycobacterial strains and increase diagnostic sensitivity. Fourth, while tNGS provides robust species identification, its high-cost, complex workflow and infrastructure to conduct the testing limits its scalability. More cost-effective molecular tools such as GenoType CM/AS line-probe assays may offer practical alternatives in resource-limited settings. A few biochemical tests such as niacin test or nitrate test can be performed in low resource settings to differentiate between MTBC and NTMs, however, limited to diagnostic capacity for species identification and differentiation. Fifth, discordant IS6110 PCR findings raise the possibility of cross-reactivity, mixed infections, or primer specificity limitations. Finally, the proposed rpoB2 Ct threshold was derived from exploratory ROC analysis and lacks external validation. Although post hoc power analysis suggested adequate statistical power to detect discriminatory performance, the study was not designed with an a priori sample size calculation specifically to validate this cutoff, and therefore the threshold should be interpreted cautiously.

Future studies should include larger and more diverse patient cohorts, integrate clinical outcome data, and evaluate the performance of alternative media and diagnostic platforms. Standardization of Ct-based thresholds for NTM screening using Xpert Ultra also warrants further validation.

This study provides the first evidence that routine Xpert Ultra testing can offer valuable insights into NTM detection in TB-endemic settings. In a potential implementation pathway, specimens reported as “MTB not detected” but showing detectable rpoB probe amplification by Xpert Ultra could be further evaluated using the proposed rpoB2 Ct threshold. Samples with Ct values ≤ 32.7 may be prioritized for additional confirmatory testing, including PCR, culture, or tNGS where available. Samples above the threshold should not exclude NTM but may require repeat testing, clinical correlation, and additional microbiological evaluation. Periodic local quality assessment and threshold revalidation may also be necessary to account for instrument variability and regional epidemiological differences. Leveraging existing diagnostic tools in this way offers a cost-effective, scalable strategy to improve NTM detection, particularly where TB and NTM burdens overlap.

Supplementary Information

Below is the link to the electronic supplementary material.

Supplementary Material 1 ^{(50.7KB, tiff)}

Supplementary Material 2 ^{(35.2KB, docx)}

Supplementary Material 3 ^{(28.1KB, docx)}

Author contributions

Afsana Akter Rupa: Formal analysis, Investigation, Methodology, Validation, Writing – original draft. Sk Nazmul Kabir: Investigation, Methodology, Validation, review & editing. Rumana Nasrin: Formal analysis, Investigation, Methodology, Validation, review & editing. Ashabul Islam: Formal analysis, Methodology, review & editing. S. M. Mazidur Rahman: Formal analysis, Methodology, review and editing. Ahammad Shafiq Sikder Adel: Methodology, review & editing. Andrea Maurizio Cabibbe: Formal analysis, review & editing. Arash Ghodousi: Formal analysis, review & editing. Daniela Brites: Review & editing. Mohammad Khaja Mafij Uddin: Conceptualization, Formal analysis, Investigation, Supervision, Validation, Writing – review & editing. Sayera Banu: Conceptualization, Project administration, Resources, Validation, Supervision, Writing – review & editing.

Funding

This research received no specific grant from funding agencies in the public, commercial, or not-for-profit sectors. Data has been generated as part of the routine diagnosis at TB Screening and Treatment Centers of icddr, b.

Data availability

The raw sequencing data generated during the current study are available in the NCBI Sequence Read Archive (SRA) under BioProject accession number PRJNA1400021. All data analyzed during this study are included in the main manuscript and its supplementary files.

Declarations

Competing interests

The authors declare no competing interests.