Rapid Triage of Smear-Positive Tuberculosis Suspects with a Commercial DNA Microarray: A Real-World Experience from 1100 Sputum Specimens

Abstract

Objective

To evaluate the real-world utility of a commercial DNA microarray assay for early triage of Mycobacterium tuberculosis complex (MTBC) in smear-positive sputum specimens.

Methods

This retrospective single-center study analyzed 1100 smear-positive sputum specimens submitted to Guangzhou Chest Hospital between January 2019 and December 2023. All specimens were processed in parallel using MGIT960 liquid culture and DNA microarray testing. For a subset of MGIT960 culture-positive specimens, MPT64 antigen testing was performed as per routine clinical practice for species identification. The study initially described the distribution of species identification results by microarray. Subsequently, in a subset of specimens with both microarray and MPT64 results (n=68), the agreement between the two assays in MTBC identification was evaluated using MPT64 as the reference standard.

Results

The DNA microarray provided species-level identification of mycobacteria in the smear-positive specimens, with MTBC detected in 41.4% (455/1100) of cases. In the comparative analysis with MPT64 results, the microarray demonstrated 100% sensitivity (95% CI: 83.2–100.0), 97.9% specificity (95% CI: 88.7–99.9), and a Cohen’s kappa coefficient of 0.97 for MTBC identification.

Conclusion

In a real-world setting, the DNA microarray assay demonstrates high reliability for MTBC identification while simultaneously providing species-level information for non-tuberculous mycobacteria (NTM). This capability allows the microarray assay to serve as an effective triage tool for smear-positive sputum specimens, offering species-level differentiation between tuberculosis and NTM infections, with the potential to reduce reliance on additional phenotypic confirmation pending further validation.

Introduction

Pulmonary tuberculosis remains a major global public health threat,¹ and patients with smear-positive pulmonary tuberculosis are of particular concern due to their high transmissibility, making them a critical target for epidemic control.² For these individuals, clinicians face a critical clinical dilemma in deciding promptly whether to proceed with empirical anti-tuberculosis treatment before definitive microbiological confirmation becomes available. However, acid-fast bacilli (AFB) smear microscopy cannot differentiate between the Mycobacterium tuberculosis complex (MTBC) and non-tuberculous mycobacteria (NTM). This diagnostic ambiguity is clinically consequential given the rising global incidence of NTM pulmonary disease.³ Furthermore, NTM are frequently isolated from patients with suspected tuberculosis in high-burden settings,⁴ and their clinical and radiological presentations closely mimic those of pulmonary tuberculosis.⁵ Because NTM infections closely resemble pulmonary tuberculosis, they are often misdiagnosed and subjected to empirical anti-tuberculosis therapy before microbiological confirmation.⁶ Such non-targeted treatment, irrespective of the underlying mycobacterial species, not only prolongs the window for infection control but also exposes patients to unnecessary drug toxicity and delayed initiation of appropriate management.

Currently, laboratory identification of mycobacteria primarily relies on post-culture analysis. Although liquid culture systems such as MGIT960 have improved detection rates, they still require several weeks for results.⁷ In this context, MPT64 antigen testing, owing to its high specificity, is widely used to rapidly confirm MTBC following culture positivity.⁸ Nevertheless, this assay provides only a binary result, namely MTBC-positive or “non-MTBC”, and offers no further information to guide the nuanced management of NTM infections. Consequently, the conventional workflow of culture followed by phenotypic identification inherently delays the critical step of early MTBC confirmation, thereby hindering the timely initiation of appropriate treatment and infection control measures.

While molecular assays such as GeneXpert MTB/RIF Ultra have substantially improved MTBC detection and are recommended by the WHO as a frontline test due to their high sensitivity for detecting MTBC and rifampicin resistance,⁹ they lack the capacity for species-level identification of NTM and may occasionally misclassify NTM as MTBC.¹⁰ This inherent limitation of Xpert Ultra creates a diagnostic gap for NTM speciation, particularly when MTBC is not detected but NTM infection remains clinically suspected.¹¹ DNA microarray technology, commonly referred to as the gene chip, is a high-throughput molecular diagnostic platform that addresses this specific gap by enabling simultaneous detection and species-level differentiation of multiple NTM alongside MTBC in a single assay. It enables rapid, parallel detection of specific genetic targets through hybridization between labeled nucleic acids from clinical specimens and immobilized oligonucleotide probes.¹² Commercially available DNA microarray assays can simultaneously report signals for both MTBC and multiple NTM species in a single test.¹³ Thus, the microarray complements Xpert Ultra by providing the species-level NTM identification that Xpert Ultra lacks. In high-volume tuberculosis laboratories, the primary value of this technology lies in advancing the clinical differentiation between MTBC and “non-MTBC”, thus enabling earlier clinical decisions well before culture results are available. However, existing research has largely been confined to analytical validation under controlled conditions.¹⁴ A key unanswered question is how the integration of this technology into routine diagnostic workflows concretely boosts triage efficiency in real-world settings.

To address this gap, we conducted a retrospective study to evaluate the real-world performance of a commercial DNA microarray assay in a cohort of 1100 smear-positive sputum specimens. Importantly, this study was not designed as a head-to-head comparison of technical performance against another laboratory method. Rather, it focuses on the practical role of the assay within the existing diagnostic workflow. Specifically, we assessed the reliability of the microarray in enabling rapid clinical differentiation between MTBC and “non-MTBC”, and further analyzed its impact on the current diagnostic pathway. Of particular interest was its ability to deliver species-level results at an earlier stage, potentially reducing the need for subsequent confirmatory testing such as MPT64 antigen detection. Through this real-world experience, we aim to provide actionable evidence to support the optimization of triage protocols in high-burden tuberculosis settings, enabling earlier, pathogen-informed clinical decisions in routine practice.

Materials and Methods

Specimen Collection

This retrospective analysis utilized sputum specimens submitted to the clinical microbiology laboratory of Guangzhou Chest Hospital for routine mycobacterial testing between January 2019 and December 2023. The inclusion of specimens in this analysis was based on routine laboratory criteria: (a) physicians’ requests for mycobacterial culture and identification, and (b) acid-fast bacilli positivity on initial smear microscopy. This approach mirrors the real-world diagnostic workflow encountered in routine clinical practice during the study period. Patients were instructed to provide early-morning sputum via deep expectoration after oral rinsing with sterile water. For those unable to produce spontaneous sputum, induction with hypertonic saline nebulization was performed. All specimens were collected in sterile, wide-mouth containers and transported to the laboratory within 2 hours under refrigerated conditions (4 °C). Upon laboratory receipt, samples underwent immediate processing to mitigate DNA degradation and maintain mycobacterial viability.

Sputum Smear Microscopy

Sputum smears were prepared by spreading representative portions of each specimen, such as purulent or mucopurulent material, onto clean glass slides to form thin films, which were then air-dried and heat-fixed by rapid flaming. Modified acid-fast staining was performed as follows: slides were flooded with 0.8% basic fuchsin (Baso, Zhuhai, China) and heat-dried over a flame, rinsed with distilled water, decolorized with 5% acid-alcohol until the runoff was clear, and counterstained with 0.06% methylene blue for 30 seconds. After final rinsing and air-drying, smears were examined microscopically. At least 300 oil-immersion fields were screened per slide. AFB were identified by their red coloration against a blue background. AFB positivity was graded according to WHO criteria (Table S2).¹⁵ Each staining batch included positive (AFB-positive sputum) and negative (sterile saline) controls to ensure assay reliability.

DNA Microarray Assay

Species identification was performed using the commercially available and standardized DNA microarray kit (CapitalBio Corp., Chengdu, China), strictly according to the manufacturer’s instructions. This assay is designed to directly detect and differentiate nucleic acids from 17 clinically prevalent mycobacterial species and groups from processed sputum samples. Briefly, the procedure was as follows: DNA was extracted from decontaminated sputum samples using the Lab-Aid 824 system (Zeesan Biotech, Xiamen, China). Subsequently, a single-tube, asymmetric multiplex PCR targeting the IS6110 element (for MTBC) and the hsp65 gene (for NTM) was performed using the kit-provided, fluorescently labeled primers and pre-optimized cycling conditions. The fluorescently labeled amplicons were then hybridized to species-specific probes pre-immobilized on the microarray chip. Fluorescence signals were acquired using a LuxScan 10K/B microarray scanner (CapitalBio Corp., Chengdu, China) and automatically analyzed by the manufacturer’s proprietary software. The final output of the assay was an automated report specifying the detected mycobacterial species (or complex). These raw, species-level results served as the primary data for all subsequent analyses in this study. Quality control was performed as per the kit protocol.

Liquid Mycobacterial Culture

Mycobacterial culture was performed using the BACTEC MGIT960 automated liquid culture system (Becton Dickinson, Sparks, MD, USA). Upon arrival at the laboratory, sputum specimens underwent standardized decontamination and concentration using the N-acetyl-L-cysteine-sodium hydroxide (NALC-NaOH) method. Briefly, an equal volume of 2% NALC-NaOH solution was added to each specimen, vortexed for 20 seconds, and incubated at room temperature for 15 minutes. The mixture was then diluted with phosphate-buffered saline (PBS, pH 6.8), centrifuged at 3000 × g for 15 minutes, and the resulting pellet was resuspended in sterile PBS. A 0.5-mL aliquot of the suspension was inoculated into MGIT960 tubes supplemented with OADC enrichment and PANTA antibiotic mixture per the manufacturer’s instructions.¹⁶ Tubes were incubated at 37 °C for up to 42 days, with automated fluorescence monitoring for mycobacterial growth. The MGIT960 system provided binary results (positive or negative for mycobacterial growth) but did not differentiate MTBC from NTM. Each batch included positive (H37Rv ATCC 27294) and negative (sterile PBS) controls to ensure assay reliability.

MPT64 Antigen Test

In the routine diagnostic workflow of our laboratory during the study period, culture-positive MGIT960 specimens could undergo further differentiation using the MPT64 antigen test. MPT64 testing was not performed systematically on all MGIT960 culture-positive specimens. Instead, it was ordered solely at the discretion of the attending clinician based on individual patient management needs during routine practice. Consequently, MPT64 results were available only for a subset of specimens, reflecting real-world selective testing rather than a prospective validation design. When performed, testing utilized a commercial colloidal gold-based immunochromatographic assay (Abbott, Republic of Korea) strictly according to the manufacturer’s instructions. Briefly, 100 μL of supernatant from MGIT960-positive cultures was applied to the test cassette. Results were interpreted after 15 minutes based on the presence of specific bands. A positive MPT64 result was interpreted as confirmation of MTBC. A negative result indicated the absence of detectable MTBC antigens and was operationally classified as “non-MTBC”, a category that may include NTM, other acid-fast organisms, or rare MTBC strains with mpt64 gene alterations. Notably, MPT64 testing was not performed on all MGIT960-positive specimens due to its selective use in routine practice. Therefore, MPT64 results were available for only a subset of specimens, which were used as the reference standard for concordance analysis with DNA microarray results.

Case Definitions and Triage Workflow

In our routine diagnostic workflow, the DNA microarray assay was performed directly on smear-positive sputum specimens. For the purpose of clinical triage, DNA microarray species-level identifications were mapped to a binary classification aligned with MPT64: MTBC identifications were assigned to the MTBC category, and NTM identifications to the “non-MTBC” category. These results guided initial management decisions pending final culture confirmation.

Concordance Analysis Between Microarray and MPT64

To assess concordance between the DNA microarray assay and the MPT64 antigen test, we analyzed a subset of specimens with paired results from both methods. A binary classification scheme (MTBC or “non-MTBC”) was applied to align the microarray output with the MPT64 antigen test, which served as the reference standard for MTBC identification. Based on this classification, a 2×2 contingency table was constructed to calculate diagnostic agreement metrics, including sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and Cohen’s kappa statistic. Given that PPV and NPV are highly dependent on the prevalence of MTBC within the analytic subset, only point estimates were reported to acknowledge this contextual limitation. In contrast, sensitivity, specificity, and the kappa statistic were presented with 95% confidence intervals, with those for sensitivity and specificity calculated using the Wilson score method and that for the kappa statistic using the Fleiss normal approximation. All statistical analyses were performed using SPSS version 22.0 (SPSS, Chicago, IL, USA).

Results

Characteristics of the Study Population

This study included 1100 patients with smear-positive sputum who were suspected of having tuberculosis (Table 1 and S1). The age distribution was predominantly middle-aged and elderly, with 41.3% aged 45–64 years, 38.8% aged ≥65 years, 15.4% aged 25–44 years, and 4.5% aged ≤24 years. The study population comprised 57.8% male and 42.2% female patients. All specimens were sputum samples subjected to MGIT960 culture, of which 1005 (91.4%) were culture-positive. All specimens were smear-positive, with a distribution of positivity grades as follows: “1–9 bacilli” accounted for 5.4%, 1+ for 27.8%, 2+ for 26.3%, 3+ for 25.8%, and 4+ for 14.7%. These findings reflect the typical epidemiology of tuberculosis, with higher prevalence among middle-aged and elderly individuals and a slight male predominance.

Table 1.

Demographic and Clinical Characteristics (n=1100)

Characteristics	n	%
Age
≤24	50	4.5
25–44	169	15.4
45–64	454	41.3
≥65	427	38.8
Gender
Female	464	42.2
Male	636	57.8
Sample type
Sputum	1100	100
MGIT960-positive	1005	91.4
Smear-positive	1100	100
1–9 bacilli	59	5.4
1+	306	27.8
2+	289	26.3
3+	284	25.8
4+	162	14.7

Species Distribution Identified by DNA Microarray

Among the 1100 smear-positive specimens, DNA microarray analysis identified 455 (41.4%) as MTBC (Table 2 and S1), while a notable proportion of the remaining specimens yielded NTM. The predominant complexes identified were Mycobacterium avium-intracellulare complex (MAC) in 156 cases (14.2%) and Mycobacterium chelonae-abscessus complex (MCAC) in 72 cases (6.5%). Among single-species NTM identifications, M. intracellulare (n=53), M. avium (n=35), and M. kansasii (n=24) were the most frequent. Notably, 17 cases of mixed infection were identified, including co‑detection of MTBC with MAC, MCAC, or single NTM species, highlighting the clinical importance of considering possible co‑infection. Additionally, 16 specimens yielded only a genus‑level signal (Mycobacterium spp.) without further classification, and 264 (24.0%) showed no detectable mycobacterial signal, which may be attributable to sample degradation, PCR inhibition, or low bacterial load. These results demonstrate that the microarray not only efficiently identifies MTBC but also simultaneously provides precise classification of NTM, thereby offering a basis for differentiated clinical management.

Table 2.

Mycobacterium Species Identified by DNA Microarray Analysis

Mycobacterium Species	Total (n=1100)
MTBC	455
MAC	156
MCAC	72
M. intracellulare	53
M. avium	35
M. kansasii	24
M. fortuitum	4
M. gordonae	2
M. scrofulaceum	1
M. szulgai	1
MTBC+MAC	7
MTBC+MCAC	2
MTBC+M. intracellulare	3
MTBC+M. avium	2
MTBC+M. kansasii	1
MTBC+M. terrae	1
MCAC+M. intracellulare	1
M. spp.a	16
Negative	264

Notes: ^aSignal detected at genus level but insufficient for MTBC/NTM classification; +, Mycobacterium coinfections.

Abbreviations: MTBC, Mycobacterium tuberculosis complex; MAC, Mycobacterium avium-intracellulare complex; MCAC, Mycobacterium chelonae-abscessus complex; M. spp., Mycobacterium species.

Real-World Diagnostic Pathway and Implications for Triage

To illustrate the actual diagnostic pathway for the 1100 smear-positive sputum specimens in this real-world clinical setting, a flow diagram was constructed (Figure 1). All specimens underwent DNA microarray testing and MGIT960 culture in parallel. Of the 1005 culture-positive specimens, MPT64 antigen testing was performed on only 68 (6.8%) upon clinical request. The remaining 937 (93.2%) were triaged directly based on DNA microarray species identification results. This workflow highlights that, in current practice, microarray results are routinely used to guide early clinical triage. The subset of 68 specimens with both microarray and MPT64 results provided a valuable opportunity to assess concordance between the two methods (Table 3).

Figure 1.

Real-world diagnostic triage workflow for mycobacterial differentiation in smear-positive sputum specimens.

Table 3.

Concordance Between DNA Microarray and MPT64 for MTBC Identification in a Subset of 68 Smear- and Culture-Positive Sputum Specimens

Method	DNA Microarray
	MTBC	Non-MTBC
MPT64
MTBC	20	0
Non-MTBC	1	47
Diagnostic metrics	Value	95% CI
Sensitivity	100%	83.2–100.0
Specificity	97.9%	88.7–99.9
PPVa	95.2%	—
NPV	100%	—
Kappa	0.97	0.88–1.00

Notes: ^aPositive and negative predictive values are highly dependent on disease prevalence. The PPV and NPV values reported here are specific to this selected sub-cohort of 68 smear- and culture-positive specimens, in which the prevalence of MTBC was 29.4% (20/68). These results should not be generalized to other patient populations with different MTBC proportions. Therefore, 95% confidence intervals for PPV and NPV are not presented.

Abbreviations: MTBC, Mycobacterium tuberculosis complex; PPV, positive predictive value; NPV, negative predictive value; CI, confidence interval.

Concordance Analysis Between DNA Microarray and MPT64 Detection

Before evaluating classification agreement with MPT64, we first assessed the overall detection concordance between the DNA microarray and MGIT960 liquid culture across all 1100 smear-positive specimens (Table S3). The microarray demonstrated a sensitivity of 77.2% (95% CI: 74.5–79.8%) and specificity of 36.8% (95% CI: 27.2–47.4%) for detecting mycobacteria relative to culture. Among the 264 specimens (24.0%) with negative microarray results, 229 (86.7%) were culture-positive. Conversely, 60 specimens were microarray-positive but culture-negative. To evaluate concordance between the DNA microarray assay and MPT64 antigen testing, we analyzed 68 sputum specimens that were smear-positive and MGIT960 culture-positive, and for which results from both the microarray and MPT64 were available (Table 3). Using the MPT64 result as the reference standard, the DNA microarray demonstrated a sensitivity of 100% (95% CI: 83.2–100.0) and a specificity of 97.9% (95% CI: 88.7–99.9) for identifying MTBC. Agreement between the two methods was substantial, with a Cohen’s kappa of 0.97 (95% CI: 0.88–1.00). In this specific cohort, the positive predictive value (PPV) and negative predictive value (NPV) were 95.2% and 100%, respectively. It should be noted that predictive values are highly dependent on disease prevalence; therefore, the PPV and NPV reported here apply specifically to this smear- and culture-positive subset, which has a high proportion of MTBC cases, and should not be generalized to other patient populations with substantially different tuberculosis prevalence.

Discussion

Based on real-world data from 1100 smear-positive sputum specimens, this study evaluated the utility of a commercial DNA microarray assay for early triage. The assay provided direct species identification results from smear-positive sputum. In the subset of specimens with available MPT64 results (n = 68), using MPT64 as the reference standard, the microarray demonstrated 100% sensitivity, 97.9% specificity, and a kappa value of 0.97 for identifying MTBC, indicating a high level of concordance between the two methods. These findings support the use of the microarray as a reliable tool for early differentiation between MTBC and “non-MTBC” in smear-positive tuberculosis suspects, providing a basis for clinical triage.

The high concordance (kappa = 0.97) and 100% sensitivity for MTBC detection demonstrated by the DNA microarray in comparison with MPT64 in our study are consistent with the performance of this technology previously reported under controlled conditions. An early validation study using clinical isolates reported that the microarray exhibited 99.6% sensitivity and 100% specificity for MTBC identification compared to sequencing.¹⁴ However, real-world evidence is needed to determine whether the performance of this technology, established under controlled conditions, can be consistently reproduced in the routine, high-throughput setting of a clinical laboratory. Existing real-world evidence has primarily focused on applying the real-time PCR technology of the same platform (CapitalBio) to diverse sample types, such as cerebrospinal fluid,¹⁷ smear-negative respiratory specimens,¹⁸ and bone tissue.¹⁹ In these contexts, its sensitivity has been shown to be comparable to that of Xpert MTB/RIF, offering a valuable alternative for mycobacterial detection. However, data on the real-world application of this specific DNA microarray in the core scenario of smear-positive sputum specimens have remained limited prior to this report. Our study evaluated the performance of this DNA microarray within the routine diagnostic pathway for smear-positive sputum specimens. All testing was conducted following standard operating procedures, and results were directly used for MTBC triage. The transition from controlled validation to routine implementation demonstrates that the technology maintains high analytical performance under real-world conditions, supporting its integration into standard diagnostic workflows.

Beyond the identification of MTBC, this DNA microarray assay provides species-level identification of NTM directly from smear-positive sputum. As shown in Table 2, it detected clinically significant NTM species, such as MAC, MCAC, and M. intracellulare. This capability complements MPT64 antigen testing, which is designed to provide a timely binary distinction between MTBC and “non-MTBC”. Accurate identification of NTM species is a prerequisite for individualized clinical management, as pathogenicity, disease progression, and antibiotic susceptibility profiles vary considerably among different NTM species.²⁰ For instance, MAC typically shows good in vitro susceptibility to clarithromycin and amikacin,²¹ whereas MCAC is often resistant to many conventional agents but may retain activity against amikacin and tigecycline.²² Consequently, misdiagnosing NTM infection as tuberculosis can lead to inappropriate therapy and delayed treatment. Currently, precise identification of NTM relies primarily on two technological platforms: molecular sequencing and matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS). Multi-gene sequencing using various combinations of loci such as 16S rRNA, hsp65, rpoB, argH, and cya provides high species-level resolution.^23–25 MALDI-TOF MS, which is based on protein profiling, enables rapid and cost-effective identification of a large number of clinical isolates. However, its accuracy depends critically on the completeness of reference databases and the performance of analytical algorithms.²⁶ Regular database updates can therefore substantially improve the reliability of species identification.²⁷ Recently, emerging nucleotide MALDI-TOF MS approaches have demonstrated potential for rapid, direct species-level identification from respiratory specimens.²⁸ In contrast, the MPT64 antigen test yields only a binary classification of MTBC versus “non-MTBC”. The DNA microarray, by providing species-level identification of NTM directly from smear-positive specimens, enables earlier microbiological differentiation. This capability supports timely clinical triage, particularly in settings where NTM prevalence is substantial.

In this study, we observed one discordant specimen that was MGIT960 culture-positive and identified as MTBC by the DNA microarray, but tested negative by the MPT64 antigen test (Table 3). In light of previous literature, this discrepancy most likely stems from mutations in the mpt64 gene leading to loss of antigen expression or alteration of epitopes, thereby compromising detection by the immunochromatographic assay. Studies have shown that false-negative MPT64 results are associated with specific genotypes, such as non-Beijing strains (mainly lineage 1 or 4) harboring a 63 bp deletion, which truncates the protein and eliminates critical antigenic epitopes.²⁹ Additionally, the lineage 5 strains prevalent in West Africa frequently carry the mpt64 128T>A (I43N) mutation, which can also substantially reduce MPT64 sensitivity.³⁰ Although whole-genome sequencing was not performed on this specimen to confirm the exact mutation, the discordant result suggests a potential cause of MPT64 false negativity. Infections caused by MTBC strains harboring genetic alterations in the mpt64 gene may be misclassified as “non-MTBC”. In contrast, the DNA microarray targets conserved sequences such as IS6110 and is unaffected by the status of the mpt64 gene, thereby demonstrating greater diagnostic reliability in such scenarios. This finding further supports the rationale for implementing the microarray as a triage tool in high-burden settings, as it not only provides NTM speciation but also effectively mitigates the risk of missed diagnoses due to MPT64 false negativity.

Discordant results between the microarray and culture require careful interpretation (Table S3). Among 229 microarray-negative/culture-positive specimens, low mycobacterial burden in low-grade smear samples,³¹ pre-analytical factors such as uneven aliquoting or PCR inhibitors,³² and potential gaps in probe coverage for rare NTM taxa likely contributed to false negatives. Conversely, the 60 microarray-positive/culture-negative cases may reflect detection of nonviable DNA from previously treated patients³³ or genotype-specific culturability limitations;³⁴ in triage practice, these results still warrant clinical evaluation rather than dismissal. Together, the whole-cohort detection concordance (Table S3) and high classification agreement with MPT64 in the 68-specimen subset (Table 3) form complementary evidence: the former establishes overall sensitivity for mycobacterial detection, while the latter confirms reliable MTBC/non-MTBC differentiation when positive. This two-tiered validation supports the microarray as a rapid triage tool, with culture remaining essential for viability confirmation and capturing low-biomass cases.

This study has several limitations that should be considered when interpreting the results. First, the MPT64 test, which served as the comparator for the concordance analysis, was performed only on a subset of MGIT960 culture-positive specimens (n = 68). The size of this subset limits the precision of the sensitivity estimate (95% CI: 83.2–100.0%), but its composition reflects the clinician-driven, selective use of MPT64 based on individual case requirements. It should be clarified that the MPT64 antigen test was selected as the reference standard not due to its absolute diagnostic accuracy, but because it represented the actual clinical practice for differentiating MTBC from “non-MTBC” in culture-positive specimens at our institution during the study period. In the routine diagnostic pathway at our institution, the MPT64 result serves as a key microbiological confirmation for smear- and culture-positive specimens, supporting final reporting and triage decisions. Therefore, the primary objective of this study was not to assess the absolute diagnostic accuracy of the microarray, but to validate its ability to reliably integrate into and potentially optimize the existing workflow. Nevertheless, it must be acknowledged that MPT64 relies on antigen expression and is susceptible to false negatives in MTBC strains harboring mpt64 gene deletions or mutations. Consequently, discordant results where the microarray identifies MTBC but MPT64 is negative may represent true MTBC cases missed by antigen testing rather than microarray false positives, highlighting that DNA-based methods like the microarray are not affected by mutations that disrupt antigen expression. Second, this was a single-center, retrospective analysis. All specimens were derived from Guangzhou Chest Hospital, which may introduce geographic and epidemiological bias, and smear-negative specimens were not included, limiting the generalizability of the findings to other populations or settings with lower tuberculosis burden. Finally, this study did not capture time-point data from specimen receipt to result reporting. Instead, based on the inherent characteristic that microarray results are obtainable shortly after smear positivity, we made a reasonable inference regarding its potential to reduce clinical decision delay.

Conclusion

This study suggests that a commercial DNA microarray assay enables direct, species-level identification of both MTBC and NTM from smear-positive sputum in a real-world setting. Based on the observed concordance with MPT64 in a limited subset, the assay may serve as an early triage tool with the potential to reduce reliance on phenotypic confirmation, though further validation is warranted. Integration of this assay into routine diagnostics could support timely etiological classification and clinical decision-making, pending prospective evaluation of turnaround time and clinical impact. Future research should include multicenter prospective studies, direct comparisons with culture-based reference methods, and cost-effectiveness analyses to comprehensively validate the generalizability and clinical utility of this triage strategy across diverse epidemiological settings.

Funding Statement

The authors declare that no external funding was provided for this research.

Ethics Approval and Consent to Participate

The study protocols received approval from the Human Ethics Committee of Guangzhou Chest Hospital (Approval No. [2022-75]) and complied with the principles outlined in the Helsinki Declaration. Informed consent was obtained from all participants through signed forms. Clinical trial number: not applicable.

Disclosure

The authors declare no commercial or financial relationships that could be interpreted as a potential conflict of interest.