Diagnostic performance of nucleic acid MALDI-TOF MS in mycobacterial detection, species identification, and drug resistance testing for Mycobacterium tuberculosis Authors: Mengdie Yan, Jing Xu, Zhengzheng Yang, Lin Liu, Yu Wang, Xiaoni Guo, Xiao Liu, Huiru An, Tianhao Wang Published: 2026 ABSTRACT BackgroundConventional methods for detecting mycobacteria were insufficient for comprehensive species identification and full-spectrum drug resistance profiling. This study aimed to evaluate the diagnostic value of nucleic acid matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) for mycobacterial detection, species identification, and Mycobacterium tuberculosis drug resistance testing.MethodsA retrospective study was performed on clinical specimens (including sputum, bronchoalveolar lavage fluid, pleural fluid, pus, and other lesion samples) obtained from patients with tuberculosis, nontuberculous mycobacterial infection, or other infectious diseases at the Eighth Medical Center of Chinese PLA General Hospital between December 2020 and July 2023. Specimens were examined using nucleic acid MALDI-TOF MS, Gene Xpert MTB/RIF (Xpert), and Mycobacterium tuberculosis culture. Diagnostic performance and agreement were assessed using Mycobacterium tuberculosis culture and composite clinical diagnosis (integrating clinical manifestations, imaging findings, and multiple laboratory tests according to WS288-2017 Diagnostic Criteria for Tuberculosis of China) as reference standards.ResultsA total of 138 specimens were included in the final analysis. Compared with Mycobacterium tuberculosis culture, nucleic acid MALDI-TOF MS exhibited relatively high analytical nucleic acid sensitivity (97.7%) for MTB detection (97.7%). However, the assay showed poor specificity of only 24.2%, with slight agreement between Mycobacterium tuberculosis culture and nucleic acid MALDI-TOF MS (Kappa = 0.151). When composite clinical diagnosis was used as the reference standard, nucleic acid MALDI-TOF MS yielded seemingly favorable diagnostic indicators (sensitivity: 98.1%, specificity: 73.3%, accuracy: 92.8%, Kappa = 0.770). Its sensitivity (98.1%) was numerically superior to that of culture (39.8%) and Xpert (73.1%), with all P < 0.001. The AUC of nucleic acid MALDI-TOF MS was 0.857, which was comparable to that of Xpert (0.870) with P = 0.721, but relatively higher than that of Mycobacterium tuberculosis culture with P < 0.001. In drug resistance prediction (n = 34), nucleic acid MALDI-TOF MS exhibited moderate accuracy for rifampicin, isoniazid, streptomycin, and fluoroquinolones (AUC range: 0.760–0.840), with rifampicin resistance sensitivity reaching 95.0%. Meanwhile, a total of 29 strains, including NTM and MTB co-infections, were identified by nucleic acid MALDI-TOF MS, which were descriptive only.ConclusionsNucleic acid MALDI-TOF MS is an auxiliary tool for the rapid diagnosis of MTB and NTM and for the preliminary screening of drug resistance. The assay exhibits moderately improved sensitivity compared with culture methods and reasonably good concordance with composite clinical diagnosis. It platform may serve as an auxiliary screening method when combined with mycobacterial culture, Xpert, and comprehensive clinical evaluation to avoid overdiagnosis. Its independent clinical application requires rigorous further validation in large-scale, multi-center cohorts before widespread routine use.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1186/s12879-026-14026-7. Introduction Tuberculosis (TB) is a worldwide public health issue that threatens human health. China is one of the countries with a high burden of TB and drug-resistant TB (DR-TB). According to the 2024 WHO Global Tuberculosis Report, the estimate of TB incidence in China in 2023 was 741,000, accounting for 6.8% of the global incidence and ranking 3rd in the world. The incidence of multidrug-resistant (MDR)/rifampicin-resistant (RR) TB patients will be approximately 29,000, accounting for 7.3% of the global incidence and ranking 4th in the world [1]. The emergence and spread of drug-resistant tuberculosis not only prolongs the treatment cycle and increases the treatment cost but also reduces the cure rate [1]. Furthermore, the global incidence of nontuberculous mycobacteria (NTM) infections is rising annually [2, 3]. Nontuberculous mycobacteriosis, which is clinically and radiologically indistinguishable from pulmonary tuberculosis (PTB), exhibits intrinsic resistance to first-line anti-tuberculosis agents, thereby introducing additional complexity to both diagnostic differentiation and therapeutic decision-making [2, 4]. Early diagnosis of tuberculosis is vital for controlling the tuberculosis epidemic. However, conventional methods are flawed with critical clinical limitations. The traditional acid-fast bacillus (AFB) is rapid, inexpensive, and widely applicable. But it suffers from low sensitivity, especially for paucibacillary samples from extrapulmonary TB patients, leading to misdiagnosis [5]. Solid culture methods (i.e., the traditional Lowenstein-Jensen (LJ) Mycobacterium tuberculosis culture) are time-consuming and take 6–8 weeks, which seriously delays the initiation of effective treatment and the control of TB transmission [6]. Even BACTEC MGIT 960 liquid culture and Modified Roche culture method (retreated clinical specimens were inoculated onto modified Roche medium and incubated for mycobacterial culture) still fail to meet the demand for timely clinical decision-making [7]. Phenotypic drug susceptibility testing (pDST), the gold standard for resistance detection, requires 2–3 months to yield results and poses biosafety risks [8]. Molecular biological detection technologies such as Gene Xpert MTB/RIF (Xpert), TB-DNA PCR-fluorescent probe method (PCR method), and loop-mediated isothermal amplification (LAMP) have been widely used in the early diagnosis of tuberculosis [9–11]. However, they are insufficient for comprehensive species identification and full-spectrum drug resistance profiling [12]. Therefore, developing rapid, sensitive, and specific detection technologies for MTB diagnosis, NTM identification, and anti-tuberculosis drug resistance testing is urgently needed in clinical practice and public health work. Nucleic acid MALDI-TOF MS is a rapid genotyping platform that detects sequence variations via mass spectrometry. By measuring these masses, it identifies mycobacterial species and drug resistance mutations [13]. Nucleic acid MALDI-TOF MS is characterized by rapid turnaround, high accuracy, and relatively cost-effectiveness per test compared with other commercial molecular diagnostic kits [13]. In recent years, it has been gradually applied to the identification of a variety of pathogenic microorganisms [14–17]. In the field of TB detection, nucleic acid MALDI-TOF MS has been successfully applied to the identification of MTB complex from cultured isolates, showing high accuracy and efficiency compared with traditional culture methods. It offers a rapid turnaround time and demonstrates high sensitivity in specimens including bronchoalveolar lavage fluid (BALF) and high sensitivity in culture-negative samples [18]. It provides broad coverage, allowing simultaneous identification of MTB and NTM species [19]. Furthermore, it can identify mixed infections and heteroresistance, providing essential support for the diagnosis and management of complex cases [20]. But the diagnostic value of nucleic acid MALDI-TOF MS in tuberculosis has not been fully evaluated. Specifically, its application in integrated mycobacterial detection (distinguishing mycobacteria from other pathogenic microorganisms), accurate species identification (differentiating from NTM), and comprehensive drug resistance testing (covering common core anti-TB drugs and their major drug-resistant mutations) has not been systematically studied. Furthermore, its diagnostic performance in clinical specimens from different lesion sites (e.g., pulmonary vs. extrapulmonary specimens, which often differ in bacterial load and specimen quality) remains to be further verified. Therefore, this study is expected to provide new insights for the rapid and accurate detection of MTB and its drug resistance, and contribute to the prevention and control of the global TB epidemic. Data and methods Study subjects Clinical specimens were collected from disease-related lesion sites, including sputum, bronchoalveolar lavage fluid, pleural fluid, pus, and others. Specimens were collected from patients diagnosed with tuberculosis, nontuberculous mycobacteriosis, or other infectious diseases. The latter group included patients initially suspected of tuberculosis but ultimately confirmed to have alternative infections such as bacterial pneumonia and fungal infection. Each patient was included with only one eligible specimen. All participants were hospitalized at the Eighth Medical Center, Chinese PLA General Hospital, between December 2020 and July 2023. The detection results of nucleic acid MALDI-TOF MS, Xpert, and Mycobacterium tuberculosis culture were retrospectively collected. This study was approved by the Medical Ethics Committee of the PLA General Hospital (Approval No.: No. S-2025-003-01). The inclusion and exclusion criteria of the study Inclusion criteria: (a) patients admitted to our hospital with clinically suspected tuberculosis/nontuberculous mycobacteriosis; (b) patients who consented to nucleic acid MALDI‑TOF MS testing.Exclusion criteria: (a) specimens with severe contamination; (b) patients who did not consent to nucleic acid MALDI‑TOF MS testing. Inclusion criteria: (a) patients admitted to our hospital with clinically suspected tuberculosis/nontuberculous mycobacteriosis; (b) patients who consented to nucleic acid MALDI‑TOF MS testing. Inclusion criteria: (a) patients admitted to our hospital with clinically suspected tuberculosis/nontuberculous mycobacteriosis; (b) patients who consented to nucleic acid MALDI‑TOF MS testing. Exclusion criteria: (a) specimens with severe contamination; (b) patients who did not consent to nucleic acid MALDI‑TOF MS testing. Exclusion criteria: (a) specimens with severe contamination; (b) patients who did not consent to nucleic acid MALDI‑TOF MS testing. Diagnostic criteria The diagnostic criteria for pulmonary Mycobacterium tuberculosis refer to WS288-2017 Diagnostic Criteria for Tuberculosis of China [21] Confirmed Diagnosis Presence of imaging findings of active pulmonary tuberculosis, with exclusion of other lung diseases, plus any one of the following: ① sputum smear positive; ② sputum Mycobacterium tuberculosis culture positive; ③ sputum molecular biology positive; ④ pulmonary histopathological examination positive.Absence of the typical imaging findings of active pulmonary tuberculosis, with exclusion of other lung diseases, plus any one of the following: ① positive sputum acid-fast staining for 2 times; ② positive sputum Mycobacterium tuberculosis culture for 1 time plus positive sputum smear for 1 time; ③ pulmonary histopathological examination positive. Presence of imaging findings of active pulmonary tuberculosis, with exclusion of other lung diseases, plus any one of the following: ① sputum smear positive; ② sputum Mycobacterium tuberculosis culture positive; ③ sputum molecular biology positive; ④ pulmonary histopathological examination positive. Presence of imaging findings of active pulmonary tuberculosis, with exclusion of other lung diseases, plus any one of the following: ① sputum smear positive; ② sputum Mycobacterium tuberculosis culture positive; ③ sputum molecular biology positive; ④ pulmonary histopathological examination positive. Absence of the typical imaging findings of active pulmonary tuberculosis, with exclusion of other lung diseases, plus any one of the following: ① positive sputum acid-fast staining for 2 times; ② positive sputum Mycobacterium tuberculosis culture for 1 time plus positive sputum smear for 1 time; ③ pulmonary histopathological examination positive. Absence of the typical imaging findings of active pulmonary tuberculosis, with exclusion of other lung diseases, plus any one of the following: ① positive sputum acid-fast staining for 2 times; ② positive sputum Mycobacterium tuberculosis culture for 1 time plus positive sputum smear for 1 time; ③ pulmonary histopathological examination positive. Composite clinical diagnosis Presence of imaging findings of suspicious active pulmonary tuberculosis, with exclusion of other lung diseases, plus any one of the following: ① suspicious tuberculosis symptoms; ② evidence of Mycobacterium tuberculosis infection; ③ positive extrapulmonary histopathological examination. In this article, all the above-mentioned diagnoses are collectively referred to as the “composite clinical diagnosis”. They were established by integrating clinical manifestations, imaging findings, and multiple laboratory tests according to China’s WS288-2017 Diagnostic Criteria for Tuberculosis. The diagnostic criteria for extrapulmonary tuberculosis refer to Technical Guidelines for Tuberculosis Prevention and Control in China (2021 Edition) [22]. The diagnosis is confirmed if either or both of the following conditions are met positive results of Mycobacterium tuberculosis etiological tests on extrapulmonary tissues, secretions, or puncture fluids.detection of typical tuberculous lesions in the local histopathological examination of the lesion. positive results of Mycobacterium tuberculosis etiological tests on extrapulmonary tissues, secretions, or puncture fluids. positive results of Mycobacterium tuberculosis etiological tests on extrapulmonary tissues, secretions, or puncture fluids. detection of typical tuberculous lesions in the local histopathological examination of the lesion. detection of typical tuberculous lesions in the local histopathological examination of the lesion. The diagnostic criteria for NTM infection refer to the Guidelines for the Diagnosis and Treatment of Nontuberculous Mycobacteriosis (2020 Edition) [2] All diagnoses are made through comprehensive analysis by integrating epidemiological history, clinical manifestations, chest imaging, and relevant auxiliary examinations, on the premise of excluding other diseases. Research methods Both Mycobacterium tuberculosis culture and Xpert were processed and detected by the Clinical Laboratory of the Tuberculosis Research Institute in the Eighth Medical Center of the PLA General Hospital, in accordance with Codes of Laboratory Tests for Tuberculosis [23]. Nucleic acid MALDI-TOF MS technology Nucleic acid MALDI-TOF MS technology is based on Shanghai Kangli Medical Laboratory, and the experimental operations from clinical specimen processing, nucleic acid extraction to mass spectrometry detection are completed by this laboratory. The main process includes: ① Multiplex PCR reaction: amplification conditions were 50 ℃, 25 min, 95 ℃, 5 min, (94 ℃, 15 s; 55 ℃, 30 s; 68 ℃, 20 s) × 45 cycles. The target fragment containing the SNP was obtained. ② Shrimp alkaline phosphatase reaction: remove excess dNTP. ③ Single-base extension reaction: add extension primer and other reactants under the reaction conditions of 94 ℃, 30 s, (94 ℃, 5 s, [52 ℃, 5 s, 80 ℃, 5 s] × 5 cycles) × 30 cycles, 72 ℃, 3 min. ④ Desalting treatment. ⑤ Mass spectrometry detection and analysis: the specimens were spotted on the SpectroCHIP chip plate and put into the nucleic acid mass spectrometer to run, and the mass spectrometry results were analyzed in Typer 4.0 software at the end of the run. Negative controls and positive controls were set for each round of testing. Mycobacterium tuberculosis culture Clinical specimens were inoculated with modified Roche medium and set at 37℃ for mycobacterial culture after pretreatment. Isolates with positive Mycobacterium tuberculosis culture were subjected to p-nitrobenzoic acid (PNB) drug-containing medium for differential culture, and negative PNB culture results were identified as MTB, and positive as NTM. Xpert GeneXpert real-time fluorescence PCR detection system and Xpert MTB/RIF detection kit (Cepheid, USA) were used. Add 1–2 ml of sputum specimen to 1–2 times the volume of SR specimen pretreatment solution provided in the kit, screw the screw cap of the treatment tube tightly, and shake the tube on a vortex shaker for at least 15–20 s until the sputum specimen is fully dissolved. Place the tube in a biosafety cabinet, and let it stand at room temperature for 15 min; open the test kit, take 2 mL of the treated specimen, and add it into the kit from the sampling hole, and then test it on the machine. Read the results according to the software prompts. Statistical methods SPSS 23.0 statistical software was used for statistical analysis. Continuous variables were described as median (interquartile range, IQR). Categorical variables were described as “number of cases (percentage)”, i.e. “(N, %)”. The difference between two groups was compared by McNemar χ² test. The diagnostic efficacy of nucleic acids MALDI⁃TOF MS and Xpert was analyzed using culture and composite clinical diagnosis as criteria. The differences in the detection rates of Mycobacterium species in different types of specimens (tissues and body fluids) were also evaluated. Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), accuracy were calculated as follows: sensitivity = \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\frac{\mathrm{TP}}{\mathrm{TP+FN}}$$\end{document} × 100%, specificity = \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\frac{\mathrm{TN}}{\mathrm{TN+FP}}$$\end{document} × 100%, PPV = \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\frac{\mathrm{TP}}{\text{}\mathrm{TP+FP}}$$\end{document} × 100%, NPV = \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\frac{\mathrm{TN}}{\mathrm{TN+FN}}$$\end{document} × 100%, accuracy = \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\frac{\mathrm{TP+TN}}{\mathrm{N}}$$\end{document} × 100%. Where TP: true positive, FN: false negative, TN: true negative, FP: false positive, N: total number of cases. For drugs with fewer than 10 cases or no resistant isolates, diagnostic performance and 95% confidence intervals (CIs) were not calculated, and only descriptive analysis was performed. For drugs with 10 or more cases, 95% CIs were estimated using the Wilson method, with a significance level of α = 0.05. Receiver operating characteristic (ROC) curve was plotted, and the area under ROC curve (AUC) was calculated. Comparison of AUCs was performed using the DeLong test. A P value < 0.05 was considered statistically significant. The agreement between test results was assessed using the Kappa coefficient (κ), calculated as \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\upkappa\:}=\frac{\mathrm{P}\mathrm{o}-\mathrm{P}\mathrm{e}\:}{1-\mathrm{P}\mathrm{e}}$$\end{document}, where Po is the observed agreement and Pe is the expected agreement by chance. The strength of agreement was interpreted as follows: 0.00–0.20: slight agreement; 0.21–0.40: fair agreement; 0.41–0.60: moderate agreement; 0.61–0.80: substantial agreement; 0.81–1.00: almost perfect agreement [24]. P < 0.05 was considered statistically significant. Results General information Patients were diagnosed with TB, NTM disease, or other infectious diseases. Specimens were classified according to the final composite clinical diagnosis of the patients, among which 111 specimens were from TB patients. The age range of the patients was 12–86 years old, and the median age was 43 years old. Females were predominant (92, 55.8%). There were 29 sputum specimens and 85 bronchoalveolar lavage fluid specimens (Table 1). Table 1Demographics and clinical disease classification of the study populationIndicatorN (%)Median Age (IQR)43 (27–58)Sex Male73 (44.2) Female92 (55.8)Specimen type Sputum29 (17.6) BALF85 (51.5) Tissue22 (13.3) Pus16 (9.7) CSF5 (3.0) Urine3 (1.8) Pleural fluid5 (3.0)Clinical Disease Classification TB111 (67.3) NTM17 (10.3) TB combined with NTM12 (7.3) Other infectious diseases25 (15.1)Abbreviations: IQR, interquartile range; BALF, bronchoalveolar lavage fluid; CSF, cerebrospinal fluid Table 1Demographics and clinical disease classification of the study populationIndicatorN (%)Median Age (IQR)43 (27–58)Sex Male73 (44.2) Female92 (55.8)Specimen type Sputum29 (17.6) BALF85 (51.5) Tissue22 (13.3) Pus16 (9.7) CSF5 (3.0) Urine3 (1.8) Pleural fluid5 (3.0)Clinical Disease Classification TB111 (67.3) NTM17 (10.3) TB combined with NTM12 (7.3) Other infectious diseases25 (15.1)Abbreviations: IQR, interquartile range; BALF, bronchoalveolar lavage fluid; CSF, cerebrospinal fluid Demographics and clinical disease classification of the study population Abbreviations: IQR, interquartile range; BALF, bronchoalveolar lavage fluid; CSF, cerebrospinal fluid Diagnostic efficacy of nucleic acid MALDI-TOF MS test for MTB Diagnostic efficacy of MALDI-TOF MS and Xpert compared with Mycobacterium tuberculosis culture Of the 165 specimens, 138 specimens were included in the final analysis to undergo MALDI-TOF MS, Mycobacterium tuberculosis culture, and Xpert testing, while the remaining 27 specimens were excluded due to incomplete clinical and laboratory data. 43 specimens were positive for mycobacterium in Mycobacterium tuberculosis culture, 114 specimens were positive for MTB by nucleic acid MALDI-TOF MS, and 79 specimens were positive for MTB by Xpert. With Mycobacterium tuberculosis culture as the standard, the sensitivity, specificity, PPV, NPV, and accuracy of nucleic acid MALDI-TOF MS were 97.7%, 24.2%, 36.8%, 95.8%, and 47.1%, respectively (Table 2). The Kappa value was 0.151, P = 0.004, indicating slight agreement between nucleic acid MALDI-TOF MS and culture. Using TB culture as the standard, the sensitivity of nucleic acid MALDI-TOF MS (97.7%, 42/43) was significantly higher than that of Xpert (88.4%, 38/43). The differences were statistically significant (McNemar χ² = 69.050, P = 0.004; McNemar χ² = 28.170, P < 0.001). ROC curves were plotted for nucleic acid MALDI-TOF MS and Xpert, and the AUCs were 0.610 (95% CI, 0.510–0.710) and 0.730 (95% CI, 0.640–0.810), respectively. The AUC of Xpert was significantly higher than that of nucleic acid MALDI-TOF MS (P = 0.042, DeLong test) (Fig. 1). Table 2Diagnostic efficacy of nucleic acid MALDI-TOF MS and Xpert for Mycobacterium tuberculosis compared with Mycobacterium tuberculosis cultureTest MethodMycobacterium tuberculosis cultureSensitivity (%)Specificity (%)Positive predictive value (%)Negative predictive value (%)Accuracy (%)Kappa valueMcNemar χ²P valueAUC (95% CI)Positive specimens (case, N)Negative specimens (case, N)Nucleic acid MALDI-TOF MS97.724.236.895.847.10.15169.0500.0040.610(0.510–0.710) Positive specimens (case, N)4272 Negative specimens (case, N)123Xpert88.456.848.191.566.70.36828.170< 0.0010.730(0.640–0.810) Positive specimens (case, N)3841 Negative specimens (case, N)554Note: McNemar’s χ² test was used for the paired comparison between each diagnostic method (Nucleic acid MALDI-TOF MS or Xpert) and the reference standard (Mycobacterium tuberculosis culture). Kappa values were interpreted as follows: 0.00–0.20, slight; 0.21–0.40, fair; 0.41–0.60, moderate; 0.61–0.80, substantial; 0.81–1.00, almost perfect agreement. AUC: area under the curve; 95% CI: 95% confidence interval Table 2Diagnostic efficacy of nucleic acid MALDI-TOF MS and Xpert for Mycobacterium tuberculosis compared with Mycobacterium tuberculosis cultureTest MethodMycobacterium tuberculosis cultureSensitivity (%)Specificity (%)Positive predictive value (%)Negative predictive value (%)Accuracy (%)Kappa valueMcNemar χ²P valueAUC (95% CI)Positive specimens (case, N)Negative specimens (case, N)Nucleic acid MALDI-TOF MS97.724.236.895.847.10.15169.0500.0040.610(0.510–0.710) Positive specimens (case, N)4272 Negative specimens (case, N)123Xpert88.456.848.191.566.70.36828.170< 0.0010.730(0.640–0.810) Positive specimens (case, N)3841 Negative specimens (case, N)554Note: McNemar’s χ² test was used for the paired comparison between each diagnostic method (Nucleic acid MALDI-TOF MS or Xpert) and the reference standard (Mycobacterium tuberculosis culture). Kappa values were interpreted as follows: 0.00–0.20, slight; 0.21–0.40, fair; 0.41–0.60, moderate; 0.61–0.80, substantial; 0.81–1.00, almost perfect agreement. AUC: area under the curve; 95% CI: 95% confidence interval Diagnostic efficacy of nucleic acid MALDI-TOF MS and Xpert for Mycobacterium tuberculosis compared with Mycobacterium tuberculosis culture Note: McNemar’s χ² test was used for the paired comparison between each diagnostic method (Nucleic acid MALDI-TOF MS or Xpert) and the reference standard (Mycobacterium tuberculosis culture). Kappa values were interpreted as follows: 0.00–0.20, slight; 0.21–0.40, fair; 0.41–0.60, moderate; 0.61–0.80, substantial; 0.81–1.00, almost perfect agreement. AUC: area under the curve; 95% CI: 95% confidence interval Fig. 1ROC curves of nucleic acid MALDI-TOF MS and Xpert compared with Mycobacterium tuberculosis culture. AUCs were compared using the DeLong test. AUC of nucleic acid MALDI-TOF MS: 0.610 (95% CI, 0.510–0.710) ; AUC of Xpert: 0.730 (95% CI, 0.640–0.810). P = 0.042 (DeLong test) ROC curves of nucleic acid MALDI-TOF MS and Xpert compared with Mycobacterium tuberculosis culture. AUCs were compared using the DeLong test. AUC of nucleic acid MALDI-TOF MS: 0.610 (95% CI, 0.510–0.710) ; AUC of Xpert: 0.730 (95% CI, 0.640–0.810). P = 0.042 (DeLong test) Diagnostic efficacy of nucleic acid MALDI-TOF MS, Mycobacterium tuberculosis culture, and Xpert compared with composite clinical diagnosis The above analyses in this study demonstrate low specificity, low overall diagnostic accuracy and slight agreement between nucleic acid MALDI-TOF MS and culture. Therefore, we further investigated the agreement between nucleic acid MALDI‑TOF MS and composite clinical diagnosis. After excluding specimens without Mycobacterium tuberculosis culture and Xpert, a total of 138 specimens were included. Among these, 108 specimens were clinically diagnosed as TB. 114 specimens were positive for MTB by nucleic acid MALDI-TOF MS, while 43 specimens by Mycobacterium tuberculosis culture, and 79 specimens by Xpert. Compared with composite clinical diagnosis, the sensitivity, specificity, PPV, NPV, and accuracy of nucleic acid MALDI-TOF MS, Mycobacterium tuberculosis culture, and Xpert are presented in the following table (Table 3). For nucleic acid MALDI-TOF MS, the sensitivity, specificity, PPV, NPV, and accuracy were 98.1%, 73.3%, 93.0%, 91.7%, and 92.8%, respectively. McNemar χ² was calculated as 3.380, with P = 0.066 ( no statistically significant difference). This indicates that the method has substantial agreement with composite clinical diagnosis results, and there is no statistically significant difference between them. The sensitivity of the nucleic acid MALDI-TOF MS was numerically superior to that of Mycobacterium tuberculosis culture and Xpert. The differences between composite clinical diagnosis and Mycobacterium tuberculosis culture or Xpert were statistically significant (McNemar χ² = 50.420, P < 0.001; McNemar χ² = 16.820, P < 0.001). Consistency analysis showed that the results of nucleic acid MALDI-TOF MS were substantially consistent with composite clinical diagnosis (Kappa = 0.770), which was relatively higher than the consistency of Mycobacterium tuberculosis culture (Kappa = 0.220, fair agreement) and the Xpert (Kappa = 0.540, moderate agreement). ROC curves were plotted for nucleic acid MALDI-TOF MS, mycobacterium culture, and Xpert. The AUC of nucleic acid MALDI-TOF MS was 0.857 (95% CI, 0.76–0.96), which was relatively higher than that of Mycobacterium tuberculosis culture (AUC = 0.699). 95% CI, 0.61–0.79; P < 0.001, DeLong test). The AUC of Xpert was 0.870 (95% CI, 0.81–0.92), which was slightly higher than that of nucleic acid MALDI-TOF MS (0.857; 95% CI, 0.76–0.96), but the difference was not statistically significant (P = 0.721, DeLong test), indicating comparable overall diagnostic accuracy between the two methods. The AUC of Xpert was also relatively higher than Mycobacterium tuberculosis culture (P < 0.001, DeLong test) (Fig. 2). Table 3Diagnostic efficacy of nucleic acid MALDI-TOF MS, Mycobacterium tuberculosis culture, and Xpert compared with composite clinical diagnosisTest MethodComposite clinical diagnosisSensitivity (%)Specificity (%)Positive predictive value (%)Negative predictive value (%)Accuracy (%)Kappa valueMcNemar χ²P valueAUC (95% CI)Positive specimens (case, N)Negative specimens (case, N)Nucleic acid MALDI-TOF MS98.173.393.091.792.80.7703.3800.0660.860(0.760–0.960) Positive specimens (case, N)1068 Negative specimens (case, N)222Mycobacterium tuberculosis culture39.8100.0100.031.652.90.22050.420< 0.0010.700(0.610–0.790) Positive specimens (case, N)430 Negative specimens (case, N)6530Xpert73.1100.0100.050.879.00.54016.820< 0.0010.866(0.810–0.920) Positive specimens (case, N)790 Negative specimens (case, N)2930Note: McNemar’s χ² test was used for the paired comparison between each diagnostic method (Nucleic acid MALDI-TOF MS, Mycobacterium tuberculosis culture or Xpert) and the reference standard (composite clinical diagnosis). Kappa values were interpreted as follows: 0.00–0.20, slight; 0.21–0.40, fair; 0.41–0.60, moderate; 0.61–0.80, substantial; 0.81–1.00, almost perfect agreement. AUC: area under the curve; 95% CI: 95% confidence interval Table 3Diagnostic efficacy of nucleic acid MALDI-TOF MS, Mycobacterium tuberculosis culture, and Xpert compared with composite clinical diagnosisTest MethodComposite clinical diagnosisSensitivity (%)Specificity (%)Positive predictive value (%)Negative predictive value (%)Accuracy (%)Kappa valueMcNemar χ²P valueAUC (95% CI)Positive specimens (case, N)Negative specimens (case, N)Nucleic acid MALDI-TOF MS98.173.393.091.792.80.7703.3800.0660.860(0.760–0.960) Positive specimens (case, N)1068 Negative specimens (case, N)222Mycobacterium tuberculosis culture39.8100.0100.031.652.90.22050.420< 0.0010.700(0.610–0.790) Positive specimens (case, N)430 Negative specimens (case, N)6530Xpert73.1100.0100.050.879.00.54016.820< 0.0010.866(0.810–0.920) Positive specimens (case, N)790 Negative specimens (case, N)2930Note: McNemar’s χ² test was used for the paired comparison between each diagnostic method (Nucleic acid MALDI-TOF MS, Mycobacterium tuberculosis culture or Xpert) and the reference standard (composite clinical diagnosis). Kappa values were interpreted as follows: 0.00–0.20, slight; 0.21–0.40, fair; 0.41–0.60, moderate; 0.61–0.80, substantial; 0.81–1.00, almost perfect agreement. AUC: area under the curve; 95% CI: 95% confidence interval Diagnostic efficacy of nucleic acid MALDI-TOF MS, Mycobacterium tuberculosis culture, and Xpert compared with composite clinical diagnosis Note: McNemar’s χ² test was used for the paired comparison between each diagnostic method (Nucleic acid MALDI-TOF MS, Mycobacterium tuberculosis culture or Xpert) and the reference standard (composite clinical diagnosis). Kappa values were interpreted as follows: 0.00–0.20, slight; 0.21–0.40, fair; 0.41–0.60, moderate; 0.61–0.80, substantial; 0.81–1.00, almost perfect agreement. AUC: area under the curve; 95% CI: 95% confidence interval Fig. 2ROC curves of nucleic acid MALDI-TOF MS, Mycobacterium tuberculosis culture, and Xpert compared with composite clinical diagnosis. AUC of nucleic acid MALDI-TOF MS: 0.857 (95% CI, 0.760–0.960); AUC of Mycobacterium tuberculosis culture: 0.699 (95% CI, 0.610–0.790); AUC of Xpert: 0.870 (95% CI, 0.810–0.920). MALDI-TOF MS vs. culture: P < 0.001; Xpert vs. culture: P < 0.001; MALDI-TOF MS vs. Xpert: P = 0.721 (DeLong test) ROC curves of nucleic acid MALDI-TOF MS, Mycobacterium tuberculosis culture, and Xpert compared with composite clinical diagnosis. AUC of nucleic acid MALDI-TOF MS: 0.857 (95% CI, 0.760–0.960); AUC of Mycobacterium tuberculosis culture: 0.699 (95% CI, 0.610–0.790); AUC of Xpert: 0.870 (95% CI, 0.810–0.920). MALDI-TOF MS vs. culture: P < 0.001; Xpert vs. culture: P < 0.001; MALDI-TOF MS vs. Xpert: P = 0.721 (DeLong test) Diagnostic efficacy of nucleic acid MALDI-TOF MS for drug resistant TB In order to assess the efficacy of MALDI-TOF MS in predicting drug resistance of MTB, 34 specimens that were positive for MTB by nucleic acid MALDI-TOF MS, culture, and Xpert were further analyzed. Phenotypic DST was performed using the minimum inhibitory concentration (MIC) method of Mycobacterium tuberculosis culture. The resistance to rifampicin, isoniazid, ethambutol, streptomycin (SM), levofloxacin, kanamycin, capreomycin, prothionamide, and sodium para-aminosalicylate was detected. Among them, 20 rifampicin-resistant strains (58.8%, 20/34), 19 isoniazid-resistant strains (55.9%, 19/34), 7 ethambutol-resistant strains (20.6%, 7/34), 19 streptomycin-resistant strains (55.9%, 19/34), and 14 levofloxacin-resistant strains (41.2%, 14/34) were identified. Using phenotypic DST as the gold standard, the indicators of nucleic acid MALDI-TOF MS in predicting drug resistance were calculated. Sensitivity, specificity, PPV, NPV, and accuracy were only calculated for anti-tuberculosis drugs with relatively large sample sizes of resistant isolates. For rifampicin resistance prediction by nucleic acid MALDI-TOF MS, the sensitivity, specificity, PPV, NPV, and accuracy were 95.0%, 57.1%, 76.0%, 88.9%, and 79.4%, respectively. For drugs with extremely small sample sizes (n = 1–2) or no resistant strains (e.g., sodium para-aminosalicylate), only descriptive data are presented, and performance metrics were not calculated to avoid misleading results (Table 4). Data for drugs with extremely small sample sizes are presented descriptively in the Supplementary Tables (Table S1). Taking Mycobacterium tuberculosis culture as the reference standard, the ROC of nucleic acid MALDI-TOF MS and Xpert for MTB drug resistance detection were shown in Fig. 3. Each drug’s diagnostic performance was evaluated independently to reflect real-world clinical scenarios. Table 4Diagnostic efficacy of nucleic acid MALDI-TOF MS for the diagnosis of drug-resistant tuberculosis compared with Mycobacterium tuberculosis culture as a standardDrugsCulture methodSensitivity*(%, 95% CI)Specificity* (%, 95% CI)Positive predictive value* (%, 95% CI)Negative predictive value*(%, 95% CI)Accuracy* (%, 95% CI)ResistantSensitiveRifampicin95.0 (75.1–99.9)57.1 (28.9–82.3)76.0 (54.9–89.5)88.9 (56.5–98.0)79.4 (62.1–90.6) Resistant196 Sensitive18Isoniazid94.7 (74.0–99.9)66.7 (38.4–87.2)78.3 (56.3–91.0)90.9 (62.3–98.4)82.4 (66.0–92.0) Resistant185 Sensitive110Ethambutol28.6 (8.2–64.1)96.3 (81.0–99.9)66.7 (20.8–94.0)83.9 (66.3–93.4)82.4 (66.0–92.0) Resistant21 Sensitive526Streptomycin68.4 (44.7–86.2)100.0(84.6–100.0)100.0(75.3–100.0)71.4 (47.8–87.9)82.4 (66.0–92.0) Resistant130 Sensitive615Levofloxacin64.3 (38.6–83.7)90.0 (69.9–97.2)81.8 (52.3–94.9)78.3 (56.3–91.0)79.4 (62.1–90.6) Resistant92 Sensitive518Note: *Calculated compared with phenotypic drug susceptibility testing. The specificity and PPV of streptomycin were initially excluded due to zero frequency, causing an unstable 95% CI, and are now fully supplemented. Data for drugs with extremely small sample sizes are shown descriptively in Supplementary Table Table 4Diagnostic efficacy of nucleic acid MALDI-TOF MS for the diagnosis of drug-resistant tuberculosis compared with Mycobacterium tuberculosis culture as a standardDrugsCulture methodSensitivity*(%, 95% CI)Specificity* (%, 95% CI)Positive predictive value* (%, 95% CI)Negative predictive value*(%, 95% CI)Accuracy* (%, 95% CI)ResistantSensitiveRifampicin95.0 (75.1–99.9)57.1 (28.9–82.3)76.0 (54.9–89.5)88.9 (56.5–98.0)79.4 (62.1–90.6) Resistant196 Sensitive18Isoniazid94.7 (74.0–99.9)66.7 (38.4–87.2)78.3 (56.3–91.0)90.9 (62.3–98.4)82.4 (66.0–92.0) Resistant185 Sensitive110Ethambutol28.6 (8.2–64.1)96.3 (81.0–99.9)66.7 (20.8–94.0)83.9 (66.3–93.4)82.4 (66.0–92.0) Resistant21 Sensitive526Streptomycin68.4 (44.7–86.2)100.0(84.6–100.0)100.0(75.3–100.0)71.4 (47.8–87.9)82.4 (66.0–92.0) Resistant130 Sensitive615Levofloxacin64.3 (38.6–83.7)90.0 (69.9–97.2)81.8 (52.3–94.9)78.3 (56.3–91.0)79.4 (62.1–90.6) Resistant92 Sensitive518Note: *Calculated compared with phenotypic drug susceptibility testing. The specificity and PPV of streptomycin were initially excluded due to zero frequency, causing an unstable 95% CI, and are now fully supplemented. Data for drugs with extremely small sample sizes are shown descriptively in Supplementary Table Diagnostic efficacy of nucleic acid MALDI-TOF MS for the diagnosis of drug-resistant tuberculosis compared with Mycobacterium tuberculosis culture as a standard Note: *Calculated compared with phenotypic drug susceptibility testing. The specificity and PPV of streptomycin were initially excluded due to zero frequency, causing an unstable 95% CI, and are now fully supplemented. Data for drugs with extremely small sample sizes are shown descriptively in Supplementary Table For the individual drug analysis (n = 34), the AUC of nucleic acid MALDI-TOF MS for rifampicin resistance detection was 0.760 (95% CI, 0.580–0.930), which was comparable to that of Xpert (0.770, 95% CI, 0.600–0.950). Nucleic acid MALDI-TOF MS exhibited moderate diagnostic accuracy for isoniazid, streptomycin, and fluoroquinolone resistance, with the AUC values of 0.810 (95% CI, 0.640–0.970), 0.840 (95% CI, 0.700–0.980) and 0.770 (95% CI, 0.600–0.950), respectively. In contrast, the diagnostic efficacy of this technique for ethambutol resistance was relatively low, with an AUC of 0.620 (95% CI, 0.370–0.880). Fig. 3ROC of nucleic acid MALDI-TOF MS and Xpert for drug resistance detection, with phenotypic drug susceptibility testing as the gold standard. (A) Rifampicin (n = 34, AUC of nucleic acid MALDI-TOF MS was 0.760 (95% CI, 0.580–0.930), which of Xpert was 0.770 (95% CI, 0.600–0.950); (B) Isoniazid (n = 34, AUC = 0.81, 95% CI, 0.640–0.970); (C) Ethambutol (n = 34, AUC = 0.620, 95% CI, 0.370–0.880); (D) Streptomycin (n = 34, AUC = 0.840, 95% CI, 0.700–0.980); (E) Fluoroquinolone (n = 34, AUC = 0.770, 95% CI, 0.600–0.950). Note: Panels A–E show statistically robust individual drug analyses (n ≥ 10 for each drug in the overall study cohort). Each drug’s diagnostic performance was evaluated independently to reflect real-world clinical scenarios ROC of nucleic acid MALDI-TOF MS and Xpert for drug resistance detection, with phenotypic drug susceptibility testing as the gold standard. (A) Rifampicin (n = 34, AUC of nucleic acid MALDI-TOF MS was 0.760 (95% CI, 0.580–0.930), which of Xpert was 0.770 (95% CI, 0.600–0.950); (B) Isoniazid (n = 34, AUC = 0.81, 95% CI, 0.640–0.970); (C) Ethambutol (n = 34, AUC = 0.620, 95% CI, 0.370–0.880); (D) Streptomycin (n = 34, AUC = 0.840, 95% CI, 0.700–0.980); (E) Fluoroquinolone (n = 34, AUC = 0.770, 95% CI, 0.600–0.950). Note: Panels A–E show statistically robust individual drug analyses (n ≥ 10 for each drug in the overall study cohort). Each drug’s diagnostic performance was evaluated independently to reflect real-world clinical scenarios Analysis of drug resistance-related gene mutations for 16 anti-tuberculosis drugs (rifampicin, isoniazid, pyrazinamide, ethambutol, fluoroquinolones, streptomycin, ethionamide, prothionamide, sodium para-aminosalicylate, amikacin, kanamycin, capreomycin, cycloserine, clofazimine, bedaquiline, and linezolid) showed that a total of 30 strains carried drug resistance mutation genes detected by nucleic acid MALDI-TOF MS, accounting for 88.2% of the positive samples. Among rifampicin-related rpoB gene mutations, the rpoB531 mutation was the most common, accounting for 53.3% (16/30). For isoniazid-related mutations, the katG315 mutation accounted for 63.3% (19/30), and the inhA-15 mutation accounted for 13.3% (4/30). No mutation genes were detected for pyrazinamide, sodium para-aminosalicylate, linezolid, bedaquiline, clofazimine, or cycloserine (Table S2). Clinical value of nucleic acid MALDI-TOF MS for the identification of NTM strains A total of 29 strains, including NTM and MTB combined with NTM, were detected by nucleic acid MALDI-TOF MS. Among them, Mycobacterium intracellulare accounted for 24.1% (7/29), Mycobacterium triviale for 10.3% (3/29), Mycobacterium massiliense for 6.9% (2/29), and Mycobacterium abscessus for 6.9% (2/29). MTB combined with Mycobacterium intracellulare, MTB combined with Mycobacterium gordonae, and MTB combined with Mycobacterium abscessus each accounted for 6.9% (2/29) (Table S3). The detection results were consistent with the final composite clinical diagnosis established for each patient based on clinical, radiological, and laboratory findings. Identification of the isolates in this study was performed by nucleic acid MALDI-TOF MS. Given that no further confirmation by molecular sequencing was conducted as the reference standard, the identified NTM species were only described descriptively. Discussion Compared with traditional Mycobacterium tuberculosis detection technologies, nucleic acid MALDI-TOF MS has the advantages of high throughput, high accuracy, and rapid speed, and is clinically applied for the identification of various pathogens [16]. Among traditional MTB etiological detection technologies, smear microscopy is simple, economical, and fast, but its sensitivity is insufficient. Culture methods (including BACTEC MGIT 960 liquid culture, modified Löwenstein-Jensen culture, and Löwenstein-Jensen solid culture) are the gold standard for MTB etiological diagnosis. However, they are time-consuming and require high-end equipment, failing to meet clinical needs for the early diagnosis and treatment of tuberculosis [25]. In 2011, the World Health Organization approved the use of rapid molecular detection methods such as Xpert for tuberculosis diagnosis in specific patient populations [26]. The study found that when using Mycobacterium tuberculosis culture as the gold standard, nucleic acid MALDI-TOF MS had the highest sensitivity, which was relatively higher than that of Xpert. This was particularly evident for specimens with relatively low detection rates, leading to a substantial reduction in missed diagnosis rates. The Kappa value in this study was 0.151, indicating slight agreement between nucleic acid MALDI‑TOF MS and the culture method. In addition, nucleic acid MALDI‑TOF MS showed relatively high sensitivity but relatively low specificity in the present study. The poor consistency might be associated with the following factors: technical principle and specimen type. First, nucleic acid MALDI-TOF MS targets mycobacterial nucleic acid sequences and can detect trace amounts of mycobacterial DNA, including residual nucleic acids from dead bacteria [27]. In contrast, culture only identifies viable bacilli. This characteristic enables high sensitivity but also leads to false-positive results and reduced specificity. Second, all specimens in this study were obtained from lesion sites. However, the matrix components of some specimens were complex, such as sputum and BALF. Some specimens contained NTM, and low homology between the detection primers and certain NTM species caused mild cross‑reactions, further lowering specificity. Therefore, we further investigated the agreement between nucleic acid MALDI‑TOF MS and composite clinical diagnosis. When composite clinical diagnosis of tuberculosis was used as the reference standard, the numerically calculated sensitivity of the nucleic acid MALDI-TOF MS assay was numerically superior to that of culture and Xpert. In terms of consistency, nucleic acid MALDI-TOF MS seemly showed the best agreement with composite clinical diagnosis results. However, this metric cannot be interpreted as robust diagnostic evidence, since composite clinical diagnosis partially relied on nucleic acid MALDI-TOF MS test results, forming circular validation. The high Kappa value (0.770) between MALDI-TOF MS and composite clinical diagnosis is artificially inflated by the above bias and should not be overinterpreted as reliable concordance evidence. From a clinical perspective, the high analytical sensitivity of nucleic acid MALDI‑TOF MS provides limited auxiliary value for the early screening of suspected tuberculosis patients to reduce missed diagnoses, yet its extremely low specificity of 24.2% against culture gold standard brings a prominent risk of false-positive results and overdiagnosis. The relatively low specificity can be offset by combined verification with mycobacterial culture, Xpert, and comprehensive clinical evaluation before clinical decision-making to avoid unnecessary overtreatment caused by false-positive results. Wang et al. tested respiratory specimens (178 bronchoalveolar lavage fluids and 36 sputum samples) from 214 patients suspected of TB. The results showed that with culture method as the gold standard, nucleic acid MALDI-TOF MS for MTB detection had a sensitivity of 92.2%, specificity of 74.1%, accuracy of 82.7%, and a Kappa value of 0.660 (good consistency). When composite clinical diagnosis was used as the reference, its sensitivity (82.5%) and accuracy (86.0%) were both higher than those of the culture method (69.2% and 78.0%, respectively), with an AUC of 0.877 (higher than 0.825 of the culture method) [27]. Li et al. evaluated the performance of nucleic acid MALDI-TOF MS using 108 samples. With clinical etiological diagnosis as the gold standard, the technology showed a sensitivity of 96.91%, specificity of 100%, and accuracy of 97.22% for MTB identification. In clinical validation, the positive rate of this technology was 72.7%, which was higher than that of Xpert MTB/RIF (63.6%) and culture (54.5%). For MTB diagnosis, its sensitivity was 72.7%, specificity was 100%, and AUC was 0.864—also higher than those of Xpert (AUC 0.818) and culture (AUC 0.773) [28]. Jiang et al. retrospectively analyzed BALF specimens from 343 patients suspected of having PTB, comparing nucleic acid MALDI-TOF MS with smear microscopy, MGIT 960 culture, and Xpert MTB/RIF using composite clinical diagnosis as the gold standard. The results showed that the sensitivity (71.5%) and negative predictive value (43.1%) of the nucleic acid MALDI-TOF MS assay were significantly higher than those of smear microscopy (22.6% and 23.2%, respectively), MGIT 960 culture (40.6% and 18.9%, respectively), and Xpert (40.8% and 27.9%, respectively). Among smear-negative BALF specimens, its sensitivity (65.1%) was still higher than that of MGIT 960 culture (25.4%) and Xpert (30.7%). The accuracy of this technology (74.3%) was higher than that of the other three methods (36.7%–51.8%), and its area under the ROC curve (AUC = 0.800) was also superior [18]. These findings highlight the clinical application value and limitations of nucleic acid MALDI‑TOF MS, and provide a reference for its rational use in the diagnosis of pulmonary tuberculosis. This study also found that in terms of drug resistance detection, nucleic acid MALDI-TOF MS showed good consistency with the pDST of Mycobacterium tuberculosis culture for the resistance detection of rifampicin, isoniazid, streptomycin, levofloxacin (fluoroquinolones), prothionamide, and sodium para-aminosalicylate. Shi et al. retrospectively analyzed samples from 201 pulmonary tuberculosis patients, comparing nucleic acid MALDI-TOF-MS with smear microscopy, MGIT 960 culture, LAMP, and Xpert. Using pDST as the gold standard for drug resistance detection, the results showed that the MTB positive detection rate of this technology was 85.57%, which was significantly higher than that of smear microscopy (28.36%), culture (43.78%), LAMP (47.26%), and Xpert (64.68%). The detection rate for rifampicin resistance was 96.72%, significantly higher than Xpert’s 81.97%. The consistency with pDST for drug resistance detection ranged from 83.0% to 98.9% [29]. The above studies all indicate that nucleic acid MALDI-TOF-MS can be used as an effective tool for the diagnosis of drug-resistant tuberculosis. Wu et al. tested 115 rifampicin-resistant and 53 rifampicin-sensitive clinical isolates of MTB. The results showed that nucleic acid MALDI-TOF MS exhibited good consistency with pDST for the detection of resistance to rifampicin, isoniazid, streptomycin, amikacin, kanamycin, ethambutol, and fluoroquinolones [13], which is consistent with the findings of the present study. In recent years, the primary and secondary drug resistance rates of MTB have been rising, with relatively higher resistance rates among retreatment patients. The main reasons include irrational initial treatment regimens, abuse of antibiotics, irregular medication intake, and poor patient compliance [30]. Xpert can be used for the rapid diagnosis of pulmonary tuberculosis and rifampicin-resistant tuberculosis, but false-positive may occur in the detection of rifampicin resistance [31]. Compared with nucleic acid MALDI-TOF MS, pDST can only determine resistance to some drugs, while Xpert can only detect rifampicin resistance-related gene mutation sites. It should also be noted that nucleic acid MALDI-TOF MS may yield results inconsistent with pDST in the detection of MTB drug resistance mutations, which is considered related to the following factors: ① A small number of strains have positive resistance genes but are sensitive in pDST, which may be associated with nonsense mutations; ② TB patients infected with both sensitive and resistant MTB strains, known as heterogeneous resistance [32–33], can affect the study results; ③ Nucleic acid MALDI-TOF MS only detects changes in drug targets caused by known gene mutations to predict drug resistance, and cannot detect resistance caused by other mechanisms, which may lead to inconsistencies between pDST results and nucleic acid MALDI-TOF MS results. Zhang et al. tested 210 clinical MTB isolates (107 sensitive and 103 resistant) using pDST as the gold standard and DNA sequencing as verification. The results showed that the technology had a sensitivity of 93.2%, specificity of 98.1%, and accuracy of 95.7% for rifampicin resistance detection, with a Kappa value of 0.91 (excellent consistency) with pDST. Among 61 isolates verified by sequencing, the consistency with sequencing results was 93.4% (57/61), and it could accurately detect 8 strains (14.3%) of heterogeneous resistance (confirmed by sequencing). These results indicate that nucleic acid MALDI-TOF-MS can rapidly and accurately detect rifampicin resistance of MTB and identify heterogeneous resistance [20]. Liu et al. compared nucleic acid MALDI-TOF MS with AFB staining, LJ culture, GeneXpert, and real-time PCR using 133 clinical samples (including bronchoalveolar lavage fluid and sputum) with composite clinical diagnosis as the gold standard. The results showed that the positive detection rate of this technology for mycobacteria was 84.96%, significantly higher than that of AFB staining (29.32%). The consistency of its detection of MTB resistance to 10 anti-tuberculosis drugs with LJ culture susceptibility results ranged from 80% to 95%, and it could also detect mutations related to heterogeneous resistance and multidrug resistance [34]. The above results indicate that nucleic acid MALDI-TOF MS can detect resistance genes of multiple drugs, which is one of its advantages. It can be used as a rapid and effective method for detecting MTB drug resistance, and has important guiding significance for the accurate formulation of treatment plans for drug-resistant tuberculosis. However, the drug resistance results are preliminary and should be interpreted with extreme caution due to the limited sample size of resistant isolates. This study also applied nucleic acid MALDI-TOF MS for mycobacterial species identification. In recent years, the incidence and prevalence of NTM diseases have shown an upward trend globally [2, 3, 35]. The imaging findings of NTM lung disease are difficult to distinguish from pulmonary tuberculosis. Strains such as the Mycobacterium abscessus complex are naturally resistant to most anti-tuberculosis drugs, posing considerable challenges to the composite clinical diagnosis and treatment of NTM lung disease. There are differences in the types of medications and treatment courses between tuberculosis and NTM diseases, as well as among different NTM diseases. Therefore, rapid and accurate etiological identification to the species or even subspecies level is of great significance for the confirmation and treatment of mycobacterial infections. However, AFB cannot distinguish MTB from NTM. Mycobacterial culture can only detect the genus Mycobacterium but cannot differentiate MTB from NTM. The p-nitrobenzoic acid/thiophene-2-carboxylic acid hydrazide (PNB/TCH) growth test is required for further differentiation. MTB cannot grow on PNB medium, while NTM can. But the method fails to distinguish some special NTM strains or co-infections of NTM and MTB. Although various molecular biology methods have been used clinically for NTM species identification, such as nucleic acid hybridization assays, line probe assays, DNA sequencing, and real-time PCR, each method has its limitations. In contrast, the nucleic acid MALDI-TOF MS assay can directly identify mycobacterial species (including MTB and NTM) from clinical samples, providing more specific etiological evidence for the rapid diagnosis and precise treatment of NTM diseases [36]. Zhu et al. tested 933 clinical isolates initially identified as NTM by the MPB64 assay. The results showed that the overall correct detection rate of this technology for all 933 isolates was 91.64% (855/933), and mixed infections were detected in 18.65% (174/933) of the samples. These results indicate that nucleic acid MALDI-TOF-MS is a potential method for identifying MTB and common clinical NTM species, with particular advantages in detecting mixed infections [19]. NTM are widely distributed in nature, mostly derived from environmental sources such as soil and water. For non-pathogenic NTM species that are prone to causing contamination, comprehensive diagnosis should fully combine other test results, and close observation of the patient’s treatment response is necessary to verify the correctness of the diagnosis. The present study has several limitations that compromise the robustness, credibility, and generalizability of its findings. First, the diagnostic performance analysis is hampered by circular validation bias. The composite clinical diagnostic criteria applied in this study incorporated nucleic acid MALDI-TOF MS results as a component of the reference standard, which affected the sensitivity and consistency estimates of the assay. Consequently, diagnostic efficacy metrics derived based on the composite clinical diagnosis cannot serve as valid core evidence to support the clinical superiority of nucleic acid MALDI-TOF MS. Second, nucleic acid MALDI-TOF MS demonstrated a low specificity and consistency against mycobacterial culture. Such poor concordance indicates that standalone use of nucleic acid MALDI-TOF MS would incur a high false-positive rate and subsequent clinical overdiagnosis, precluding its application as an independent diagnostic tool. We will strictly avoid such design defects in future relevant research. Mycobacterial culture, Xpert, and comprehensive clinical evaluation need to be considered before clinical decision-making to avoid unnecessary overtreatment. Third, as a single-center and single-regional population retrospective study with a relatively small sample size, the generalizability of our findings is limited. Future multi-center, prospective, large-sample, and cross-regional studies are needed to validate the auxiliary screening value of this technology rather than its independent diagnostic value. Fourth, the insufficient case volume limited the statistical analysis of most second-line drug susceptibility results to descriptive analyses only. Expanded patient enrollment in future studies will enable more comprehensive and quantitative statistical evaluation of relevant indicators. Fifth, several drug-resistant subgroups had small sample sizes in this cohort, which prohibited a direct assessment of the concordance between genetic mutations and phenotypic drug resistance. Further studies with enlarged resistant subgroup populations are required to clarify such genotypic-phenotypic consistency. Sixth, this study primarily focused on evaluating the diagnostic efficiency of drug susceptibility testing approaches, whereas clinical follow-up data and treatment outcomes of drug-resistant patients were not collected or analyzed. Incorporating these critical clinical endpoints in future prospective studies would substantially strengthen the clinical evidence base. Seventh, species identification in the current study was not verified using molecular sequencing as the reference standard. Standardized molecular sequencing is therefore required to further confirm the accuracy and reliability of the species identification results generated by nucleic acid MALDI-TOF MS. Collectively, nucleic acid MALDI-TOF MS presents considerable clinical application constraints. Its inherent high false-positive rate results in suboptimal diagnostic performance, and the seemingly favorable sensitivity calculated from composite clinical diagnosis is biased by circular validation, which undermines its scientific credibility. Nucleic acid MALDI-TOF MS currently has clinical application limitation that its high false-positive rate may lead to unsatisfactory diagnostic efficiency. It remains unsuitable for broad adoption in large-scale clinical screening and routine clinical settings for the time being. Conclusion Nucleic acid MALDI-TOF MS is an auxiliary tool for the rapid diagnosis of MTB and NTM and for the preliminary screening of drug resistance. The assay exhibits moderately improved sensitivity compared with culture methods and reasonably good concordance with composite clinical diagnosis. It platform may serve as an auxiliary screening method when combined with mycobacterial culture, Xpert, and comprehensive clinical evaluation to avoid overdiagnosis. Its independent clinical application requires rigorous further validation in large-scale, multi-center cohorts before widespread routine use. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1