ddPCR-based profiling of plasma EV-miRNA identifies a noninvasive biomarker panel for tuberculosis Authors: Zhuhua Wu, Jing Liang, Yuchuan Zhao, Jianming Hong, Yongping Xie, Yongzhao Liu, Shende Huang, Jingwen Lai, Xiaoyu Lai, Ran Wei, Jiawen Wang, Weilong Liu, Xunxun Chen Published: 2025 ABSTRACT AbstractBackgroundTuberculosis (TB) is a top infectious disease killer; underdiagnosis fuels its spread. Thus, finding perspective non-invasive biomarkers for early tuberculosis (TB) is very essential. Extracellular vesicles (EVs) microRNAs (miRNAs) are promising non-invasive biomarkers for tuberculosis (TB) detection.MethodsTo investigate the feasibility of early diagnosis of TB, we examined plasma-derived EV-miRNAs obtained from patients with active TB (n = 25), asymptomatic tuberculosis (aTB) (n = 25), and latent tuberculosis infection (LTBI) (n = 25). Initially, EVs were isolated from plasma samples using ultracentrifugation and were characterized via nanoparticle tracking analysis and immunoblotting for surface markers. Additionally, we characterized miRNA profiles using RNA sequencing to identify TB-specific EV-miRNAs. EV-miRNAs were further validated by droplet digital PCR (ddPCR).ResultsSeven differentially expressed miRNAs among the clinical subgroups (fold change > 2, P < 0.05) showed the most significant TB-specific expression patterns. Among these, three miRNAs exhibited significant differences. Furthermore, hsa-miR-451a was upregulated and two miRNAs (hsa-miR-1908-5p and hsa-miR-1268b) were downregulated. The optimized three miRNAs panel exhibited superior diagnostic accuracy with area under the curve (AUC) values of 0.970 (95% confidence interval [CI] 0.928–1.000) for discriminating active TB from healthy controls (HCs) and 0.971 (95% CI 0.923–1.000) for distinguishing LTBI from HCs. A combination of two miRNAs (hsa-miR-451a and hsa-miR-1268b) exhibited a good capacity to distinguish patients with aTB from HCs, with an AUC of 0.880 (95% CI 0.771–0.989).ConclusionThis multigroup cohort study established a novel EV-miRNAs signature and demonstrated its clinical utility for TB staging and early detection.Supplementary InformationThe online version contains supplementary material available at 10.1186/s40001-025-03533-z.Keywords: Tuberculosis, Extracellular vesicle, MiRNA, Biomarker, Diagnosis Introduction Tuberculosis (TB) is a significant global health challenge, with the World Health Organization (WHO) 2024 Global TB Report estimating approximately 10.8 million incident cases and 1.25 million attributable fatalities in 2023 [10]. The implementation of early diagnostic interventions and prompt therapeutic regimens is critical for achieving the TB elimination targets outlined in global public health strategies [6]. In addition, current diagnostic and screening criteria for TB may not adequately distinguish asymptomatic TB (aTB) from latent tuberculosis infection (LTBI) and active TB infection [25]. Consequently, the discovery of novel TB-specific biomarkers is imperative to advance translational research and establish robust diagnostic modalities for early disease detection. MicroRNAs (miRNAs) are evolutionarily conserved small non-coding RNAs that orchestrate post-transcriptional gene silencing and critically modulate immune homeostasis by targeting mRNA destabilization or translational repression [3]. Accumulating evidence highlights the potential of miRNAs as non-invasive diagnostic biomarkers for TB, with dysregulated expression profiles strongly correlating with Mycobacterium tuberculosis (Mtb) infection stages [27]. Our prior investigations established Simoa as a highly sensitive platform for serum miRNA quantification, while concurrently demonstrating its utility as a robust non-invasive strategy for the early detection of pulmonary tuberculosis [30]. Extracellular vesicles (EVs) are nanoscale vesicles with a diameter ranging from 30 to 150 nm, which are defined by their phospholipid bilayer structure [21]. They encapsulate bioactive cargoes such as nucleic acids (miRNA, lncRNA, mRNA, and DNA), proteins, and metabolites [12]. These vesicles serve as pivotal mediators of intercellular crosstalk and immune homeostasis modulation, underpinning their essential roles in diverse pathophysiological contexts [7, 8, 14]. Previous studies have demonstrated that EVs-mediated crosstalk with Mtb has a profound impact on TB progression by modulating host immune responses, suggesting that EVs cargo components are promising candidates for novel TB diagnostic strategies [23]. Several studies have reported the diagnostic potential of EV-miRNAs in pulmonary TB, as their lipid bilayer protection confers resistance to RNase degradation and enables prolonged stability in blood and other biofluids [13, 27]. For example, miR-185-5p and miR-423-5p in EVs have been reported to be closely correlated with TB and may serve as diagnostic biomarkers in patients with TB [16, 26]. In addition, four LTBI-specific serum EV-miRNAs (miR-450a-5p, let-7e-5p, miR-140-5p, and let-7d-5p) and five active TB-specific serum EV-miRNAs (miR-370-3p, miR-1246, miR-193b-5p, miR-2110, and miR-28-3) are potentially valuable for determining the infectious stages of Mtb [18]. Despite the great potential of miRNAs as diagnostic biomarkers, current detection platforms are restricted by technical drawbacks that impede their application in clinical settings [15]. Droplet digital PCR (ddPCR), a third-generation nucleic acid quantification technology, surpasses traditional quantitative PCR (qPCR) [17]. By detecting endpoint binary signals, ddPCR enables absolute target quantification without calibration curves and shows better resistance to PCR inhibitors, which is a crucial benefit for inhibitor-rich clinical specimens. This enables the precise detection of low-abundance miRNAs, even in samples with substantial background interference. In this study, we systematically characterized the EV-miRNA profiles in human plasma samples obtained from healthy controls (HCs) and three distinct TB cohorts, active TB, aTB, and LTBI, using next-generation RNA sequencing. Subsequently, ddPCR was employed for the direct quantification of candidate miRNAs in plasma-derived EVs samples, enabling the development of a multinomial logistic regression-based diagnostic model that effectively discriminates between disease progression stages and healthy status. Materials and methods Participants in this study and sample collection A total of 100 samples were included in the study: three active TB, three aTB, three LTBI, and three HC samples were used for the next-generation sequencing experiment, while 25 active TB, 25 aTB, 25 LTBI, and 25 HC samples were analyzed via ddPCR (Table 1). The inclusion criteria were that participants were aged ≥ 18 years and tested HIV-negative prior to enrollment. At enrollment, demographic information (sex, age, and medical history) was recorded. For participants ultimately diagnosed with TB, additional details—including TB history, cavity presence, complications, and clinical symptoms—were collected. IGRA results were recorded as well. The study was conducted in accordance with the guidelines of the Declaration of Helsinki [1]. Blood collection and preparation protocols were reviewed and approved by the Center for Tuberculosis Control of Guangdong Province. All participants provided informed consent prior to sample collection. In the present study, peripheral blood samples were collected from all participants. Subsequently, the plasma was separated by centrifugation, meticulously collected, and aliquoted before being stored at − 80 °C until further analysis. Table 1Characteristics of study participantsHC (n = 25)Active TB (n = 25)aTB (n = 25)LTBI (n = 25)Age, yr39.64 ± 9.6342.76 ± 12.4433.36 ± 8.7840.62 ± 16.65Gender (%)Male10 (40.0)17 (68.0)16 (64.0)12 (48.0)Female14 (60.0)8 (32.0)9 (36.0)14 (52.0)Typical clinical symptoms of TB (%)YesNA25 (100)0 (0)0 (0)NoNA0 (0)25 (100)25 (100)Presence of cavity (%)NA9 (36.0)6 (24.0)0 (0)Presence of complications (%)NA0 (0)0 (0)0 (0)History of TB0 (0)4 (16.0)0 (0)0 (0)IGRA (%)Positive0 (0)NANA25 (100.0)Negative25 (100.0)NANA0 (0)HC, healthy control; active TB, active Tuberculosis; aTB, Asymptomatic Tuberculosis; LTBI, Latent Tuberculosis Infection; NA, Not availableaTypical clinical symptoms of TB include cough, hemoptysis, Fever, weight loss Characteristics of study participants HC, healthy control; active TB, active Tuberculosis; aTB, Asymptomatic Tuberculosis; LTBI, Latent Tuberculosis Infection; NA, Not available aTypical clinical symptoms of TB include cough, hemoptysis, Fever, weight loss Isolation of plasma EVs by ultracentrifugation Plasma EVs were isolated by ultracentrifugation as described previously [9, 24]. Initially, defrosted plasma aliquots underwent centrifugation at 3000 × g for 15 min at 4 °C to eradicate cellular remnants. Subsequently, the supernatant was subjected to a second centrifugation step at 100,000 × g for 70 min at 4 °C to concentrate the EVs (Hitachi CP100MX). The pelleted EVs were subsequently resuspended in phosphate-buffered saline (PBS) to facilitate subsequent analytical procedures. Nanoparticle tracking analysis (NTA) The size and particle concentration of plasma EVs were assessed by NTA using a Particle Metrix ZetaView. Each sample was diluted 100- to 1000-fold in PBS. The NTA measurement duration was set to 60 s with a frame capture rate of 25 per second. To ensure reliability, triplicate readings were conducted for each sample to ascertain accurate particle analysis. Transmission electron microscopy (TEM) The morphology of the plasma EVs was visualized using TEM. Briefly, 10 μL of purified EVs suspension was deposited onto a formvar/carbon-coated grids (200 mesh) and allowed to adsorb for 1 min at room temperature. The excess liquid was carefully removed by blotting the grid edge with filter paper. Subsequently, 10 μL of 2% uranyl acetate (pH 4.5) was applied to the grid for negative staining (1 min incubation), followed by blotting to eliminate residual stain. The grids were air-dried for 5 min under ambient conditions before imaging. Observations were made using a Hitachi HT-7700 TEM operating at an accelerating voltage of 100 kV. Western blotting The protein concentration was determined using a BCA assay kit (Beyotime, China) according to the manufacturer’s protocol. An equal amount of loading buffer was mixed with 20 μg of total protein, which was denatured at 95 °C for 10 min and separated on 12% SDS-PAGE gels. Proteins were transferred onto PVDF membranes, followed by blocking with 5% non-fat milk in TBST for 1 h at room temperature. The samples were subsequently subjected to overnight incubation at 4 °C with the following primary antibodies: CD63 (Abcam, ab193349), CD81 (Abcam, ab109201), TSG101 (Abcam, ab125011), and negative control calnexin (Abcam, ab133615). Subsequently, the membranes were incubated with HRP-conjugated goat anti-rabbit IgG secondary antibody for 1 h at room temperature. The protein bands were visualized using an enhanced chemiluminescent substrate (Beyotime) and imaged using a ChemiDoc MP system (Bio-Rad). RNA extraction, sequencing, and data analysis Total RNA was isolated from EVs samples using TRIzol reagent (Thermo Fisher Scientific, USA) according to the manufacturer’s instructions. RNA integrity was verified using a NanoDrop2000 spectrophotometer (Thermo Fisher Scientific, USA). Small RNA sequencing of plasma-derived EVs RNA was performed using an Illumina NovaSeq 6000 platform (FOREVERGEN Biotech Co., Ltd., Guangzhou, China) with three biological replicates per experimental group. Sequencing libraries were prepared using the QIAseq miRNA Library Kit (Qiagen) with 50 bp single-end reads. Count data were analyzed using miRDeep2 v2.0.1.3 [4]. To analyze differentially expressed miRNAs (DEMs), quantile normalization was performed to standardize the data throughout the samples. DEMs were identified using quantile-normalized expression matrices with |log2(fold change)|≥ 1.0 and p-value < 0.05 as significance thresholds. Integrated miRNA target prediction was performed using three complementary algorithms: miRDB, miRWalk, and TargetScan. The constructed regulatory networks, consisting of target genes predicted by at least 2 databases, were visualized in Cytoscape v3.9.1 using organic layout algorithms. Subsequent functional annotation of the predicted targets was performed using the DAVID bioinformatics platform with Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis. Droplet digital PCR (ddPCR) Following small RNA sequencing, seven DEMs were selected for two-phase validation. Primary technical verification was performed using the QX200 droplet system (FOREVERGEN, Guangzhou) with the same RNA aliquots prepared for sequencing. Candidates showing consistent differential expression were subsequently advanced to large-scale validation in an independent cohort (n = 100). Reverse transcription was performed using the ddPCR EvaGreen Supermix (Bio-Rad, Hercules, CA, USA). Droplets for ddPCR were prepared according to the manufacturer’s protocol and subsequently loaded into 96-well PCR plates for analysis. The specific primers used for ddPCR are listed in Supplementary Table 1. Statistical analysis Candidate miRNAs were selected based on quantile-normalized expression profiles with stringent thresholds (|log2(fold change)|≥ 1.0 and P value < 0.05). Continuous variables were expressed as mean ± standard deviation. Non-parametric comparisons between groups were performed using the Mann–Whitney U test, with P values adjusted for multiple testing. The diagnostic efficacy of candidate miRNAs was determined via receiver operating characteristic (ROC) curve analysis, which included calculation of the area under the curve (AUC) and optimal cutoffs for sensitivity and specificity (determined by Youden’s index); these details are provided in Supplementary Table 7. Statistical significance was set at P < 0.05. Results The characterization of plasma-derived EVs Plasma-derived EVs were characterized using TEM, NTA, and western blotting. TEM revealed characteristic membrane-bound structures (Fig. 1A). Further measurements by NTA showed that the average particle size of EVs was as follows: HC: 170.47 ± 12.56 nm; active TB: 164.53 ± 1.80 nm; aTB: 170.80 ± 5.41 nm; and LTBI: 180.80 ± 9.87 nm (n = 3) (Fig. 1B, C). Western blot confirmation of EVs markers (CD63, CD81, and TSG101) in patient specimens (Fig. 1D), with combined size-morphology-protein profiling, confirmed the successful isolation of plasma-derived extracellular vesicles, meeting the EVs criteria. Identification of plasma-derived EVs. A Representative TEM images of EVs. Scale bars: 100 nm. B, C Size distribution of the isolated particles were analyzed using NTA. D Western blotting was employed to validate the presence of canonical EVs-derived markers (CD63/CD81/TSG101/calnexin) in plasma-isolated EVs-derived proteins. The full-length blots and gels are showcased in Supplementary Fig. 1. TEM, Transmission Electron Microscope; NTA, Nanoparticle tracking analysis; HC, healthy control; aTB, asymptomatic tuberculosis; active TB, active tuberculosis; LTBI, latent tuberculosis infection. Data are presented as mean ± SDs; n = 3 per group Profiling of DEMs in plasma-derived EVs To identify circulating EV-miRNAs with potential for TB diagnosis, small RNA sequencing was performed on plasma EVs isolated from three HCs, three individuals with active TB, three individuals with aTB, and three individuals with LTBI. Comparative EV-miRNA profiling identified distinct differential expression patterns across TB progression stages, with a total of 752 differentially expressed miRNAs identified (Supplementary Tables 2, 3, 4, 5). Specifically, it quantified 22, 9, and 11 significantly dysregulated miRNAs (|log2(fold change)|≥ 1.0 and P value < 0.05) in active TB, aTB, and LTBI cohorts, respectively, when benchmarked against HCs (Fig. 2A). Overlapping differential expressed miRNAs (DEmiRs) identified through comparisons between active TB vs HC, aTB vs HC, and LTBI vs HC are shown in Fig. 2B, with seven overlapping DEmiRs identified (Fig. 2C). The sequencing data were further validated by ddPCR analysis of seven miRNAs (hsa-miR-1268a, hsa-miR-1268b, hsa-miR-1-3p, hsa-miR-17-5p, hsa-miR-1908-5p, hsa-miR-432-5p, and hsa-miR-451a) with differential expression levels. As shown in Fig. 3A, among the seven miRNAs, four exhibited concordant trends between the log2 fold changes from miRNA sequencing and the relative expression levels measured by ddPCR analysis. ddPCR validation results showed that three miRNAs (hsa-miR-451a, hsa-miR-1908-5p, and hsa-miR-1268b) were significantly differentially expressed (P < 0.05). Notably, hsa-miR-1268b did not reach statistical significance, and thus no further large-sample ddPCR validation was conducted (Fig. 3B). Identification of plasma-derived ev-microRNA (miRNA) expression profiles in the active TB, aTB, LTBI, and HC groups. A Volcano plots of differential miRNAs for the active TB/HC, aTB/HC, and LTBI/HC groups (green: FC < 2.0, red: FC > 2.0). B Venn diagrams show the number of miRNAs with significantly differential expression (|logFC|> = 1, P < 0.05) between groups. C Overlapping differentially expressed miRNAs (DEmiRs) between DEmiRs (active TB vs. HC), DEmiRs (aTB vs. HC), and DEmiRs (LTBI vs. HC). FC, fold change Confirmation of identified miRNAs via ddPCR. A Integration of EV-miRNA sequencing profiles and ddPCR. Scatterplots illustrating the relationship between the fold change of seven miRNAs in EV-miRNA sequencing results and that in ddPCR results. The same batch of RNAs prepared for sequencing were utilized to perform ddPCR on miRNAs. B Expression levels of hsa-miR-451a, hsa-miR-1908-5p, and hsa-miR-1268b in the active TB, aTB, LTBI and HC groups. ns, no significance; * P < 0.05; ** P < 0.01 The diagnostic potential of candidate miRNAs evaluated through ddPCR analysis We selected three DEmiRs (hsa-miR-451a, hsa-miR-1908-5p, and hsa-miR-1268b) for validation by ddPCR in a larger cohort of plasma EVs samples. ddPCR showed that EVs-derived hsa-miR-451a was significantly upregulated, whereas EVs-derived hsa-miR-1908-5p and hsa-miR-1268b were markedly downregulated in the active TB, aTB, and LTBI groups compared with those in the HC group. However, no significant difference in hsa-miR-451a expression was observed between the aTB and HC groups (Fig. 4). To further explore the diagnostic value of miRNAs, the ROC curves of the three DEmiRs for distinguishing HCs from patients with active TB, aTB, and LTBI are shown in Fig. 5. hsa-miR-1908-5p was excluded from the ROC analysis because of insufficient discriminatory power between the aTB and HC groups. When distinguishing the active TB group from the HC group, The AUC values of EVs-derived hsa-miR-451a, hsa-miR-1908-5p, and hsa-miR-1268b were 0.770 (95% confidence interval [CI]: 0.626–0.913), 0.907 (95% CI: 0.819–0.995), and 0.667 (95% CI: 0.515–0.820), respectively. The three-miRNA panel well distinguished the active TB group from the HC group with an AUC of 0.970 (95% CI: 0.928–1.000) (Fig. 5A). When distinguishing the aTB group from the HC group, The AUC values of EVs-derived hsa-miR-451a and hsa-miR-1268b were 0.854 (95% CI: 0.747–0.962) and 0.715 (95% CI: 0.571–0.859), respectively (Fig. 5B). The result showed that the two-miRNA panel well distinguished the aTB group from the HC group with an AUC of 0.880 (95% CI: 0.771–0.989 (Fig. 5B). When distinguishing the LTBI group from the HC group, the AUC values of EVs-derived hsa-miR-451a, hsa-miR-1908-5p, and hsa-miR-1268b were 0.861 (95% CI: 0.755–0.967), 0.933 (95% CI: 0.854–1.000), and 0.757 (95% CI: 0.622–0.892), respectively. The three-miRNA panel well distinguished the LTBI group from the HC group with an AUC of 0.971 (95% CI: 0.923–1.000) (Fig. 5C). ddPCR validation of the three candidate plasma-derived EV-miRNAs (hsa-miR-451a, hsa-miR-1908-5p, and hsa-miR-1268b) in the validating cohort (n = 100). The relative abundance of the miRNAs is represented in boxplots. ns, no significance; *P < 0.05; **P < 0.01; ***P < 0.001 Diagnostic potential of candidate plasma-derived EV-miRNAs. Receiver operating characteristic (ROC) curve analysis of the three candidate plasma-derived EV-miRNAs shows that the combined diagnostic model performed well in distinguishing active TB from HC (AUC = 0.970) (A), aTB from HC (AUC = 0.880), and LTBI from HC (AUC = 0.971). hsa-miR-1908-5p was excluded from ROC analysis because of insufficient discriminatory power between aTB and HC groups (P > 0.05) Functional annotation of candidate miRNAs via pathway enrichment analysis To elucidate the potential roles of the three candidate miRNAs in TB pathogenesis, bioinformatic analysis of their target genes was conducted using KEGG pathway enrichment. The analysis identified seven significantly enriched pathways (P < 0.05), with functional annotations revealing predominant associations with cellular homeostasis and adaptive regulatory mechanisms, including cell cycle progression, synaptic vesicle trafficking, and the MAPK signaling pathway (Fig. 6A). To delineate the regulatory landscape of candidate miRNAs, we established an miRNA–mRNA interaction network through integrative bioinformatics analysis and mapped putative miRNA-target associations (Fig. 6B). Intriguingly, several genes within this network—including TGFB1, STK11, HSPB1, DUSP8, GNAI2, and FASN—have been extensively documented to play critical roles in the pathogenesis and progression of TB. miRNAs target gene prediction and analysis. A KEGG pathway enrichment analysis of predicted targets for three plasma EV-miRNA candidates. B Integrated visualization of miRNA-mRNA-pathway interactions, with orange nodes representing miRNAs, green nodes indicating target mRNAs, and blue nodes denoting enriched KEGG pathways. Solid connections illustrate miRNA-target gene relationships, while dashed lines highlight associations between target genes and enriched KEGG pathways Discussion Advancements in non-invasive, rapid, and economical diagnostic techniques utilizing circulating biomarkers could offer a viable approach for TB detection. In this study, we focused on plasma EVs samples to detect DEMs between the active TB, aTB, LTBI, and HC groups using RNA sequencing. We also evaluated the diagnostic potential of the candidate plasma-derived EV-miRNAs as TB biomarkers using ddPCR. Our study represents a screening and validation of serum EV-miRNAs in aTB patients. This multigroup cohort study established a novel EV-miRNA signature and demonstrated its clinical utility for TB staging and early detection. Recent studies have highlighted the potential role of EV-miRNAs as diagnostic biomarkers of TB [19]. For example, in a comparative analysis of plasma biomarkers between 20 TB patients and 17 HCs, Kaushik et al. verified miR-185-5p as a promising diagnostic biomarker for TB, demonstrating 50% sensitivity and 93.75% specificity (AUC = 0.750) [16]. Alipoor et al. demonstrated the potential of a combination of serum EVs-derived miR-96, miR-484, and miR-425 to distinguish TB patients from HCs [2]. Additionally, emerging evidence suggests that EV-miRNAs exhibit discriminative capacity in distinguishing TB from pulmonary comorbidities such as adenocarcinoma of the lung and other benign lesions, with differential expression patterns providing potential inter-disease demarcation [11, 34]. According to our RNA sequencing and ddPCR validation, hsa-miR-451a was upregulated in the plasma EVs of active TB patients compared with that in the HC group, whereas Wang et al. discovered that hsa-miR-451a was downregulated in TB pleural effusion compared with that in non-TB benign pleural effusion (NPE) [28]. In serum samples, the expression levels of hsa-miR-451a were significantly lower in TB patients than in the HC group [32], which may be attributed to differences in specimen sources. Our data showed that plasma EVs-derived hsa-miR-1908-5p and hsa-miR-1268b levels were decreased in TB patients. However, the roles of hsa-miR-1908-5p and hsa-miR-1268b in TB have not yet been revealed. Notably, miR-1908-5p has been mechanistically linked to autoimmune pathogenesis through its capacity to disrupt B cell tolerance checkpoints [29]. The targeting of TGFB1 and DUSP8 by miR-1908-5p suggests a mechanism for modulating macrophage polarization, as TGF-β signaling is critical for balancing pro- and anti-inflammatory responses during chronic infection [31]. In our study, we did not delineate the function of each miRNA. Further investigations are required to assess the role of each miRNA in TB development. Because of the low abundance of miRNAs in plasma-derived EVs [22], we employed ddPCR, an ultrasensitive detection method, to analyze serum-derived EV-miRNAs. Our ddPCR validation strategy addressed the critical limitations of conventional qPCR in quantifying low-abundance EV-miRNAs. By partitioning reactions into 20,000 droplets, ddPCR achieved absolute quantification without relying on external standards, overcoming variability from PCR inhibitors in plasma, which is a persistent challenge in clinical miRNA studies. Recent comparative analyses have demonstrated that ddPCR exhibits enhanced diagnostic sensitivity over GeneXpert MTB/RIF and mNGS for detecting Mtb infections [33]. Recent reports have emphasized that miRNA panels made up of combined miRNAs hold great promise as biomarkers for prognosis and diagnosis [5, 19, 20]. Multiplex miRNA profiling generates enhanced diagnostic specificity through integrated biomarker signatures, enabling superior differentiation of patients with TB from HCs, compared to single-marker analysis [30]. In this study, we successfully established an miRNA panel composed of three plasma EV-miRNAs (hsa-miR-451a, hsa-miR-1908-5p, and hsa-miR-1268b) exhibiting superior diagnostic accuracy, with AUC values of 0.970 (95% CI 0.928–1.000) for discriminating patients with active TB from HCs and 0.971 (95% CI 0.923–1.000) for distinguishing patients with LTBI from HCs. A combination of two miRNAs (hsa-miR-451a and hsa-miR-1268b) exhibited a good capacity to distinguish patients with aTB from HCs, with an AUC of 0.880 (95% CI 0.771–0.989). In brief, the distinct expression patterns of miRNAs in plasma EV-miRNAs of patients with TB could offer a fresh outlook for the diagnosis of TB. Although this study advances our understanding of EV-miRNA biomarkers in TB detection, certain limitations must be addressed to strengthen clinical translatability. First, the study had a limited plasma sample size and lacked power calculation for sample size determination; although it still revealed statistically significant differences in miRNA expression, the results require multicenter validation with larger cohorts to confirm generalizability. Second, the diagnostic specificity of these miRNAs is potentially compromised by their overlapping expression profiles across various pulmonary pathologies, including bacterial pneumonia and lung neoplasms, necessitating cross-pathological validation to establish their clinical applicability, despite the observed statistical significance. Finally, systematic functional validation through mechanistic studies is required to delineate the precise regulatory mechanisms of the candidate miRNAs in mycobacterial pathogenesis and disease progression. In conclusion, our findings establish plasma EV-miRNAs as promising diagnostic biomarkers for TB, with ddPCR demonstrating particular clinical utility through its absolute quantification capability and superior sensitivity, without requiring standardized curves. ddPCR technology merits in-depth investigation in a large, comprehensively characterized, longitudinal patient cohort. Supplementary Information Additional file 1 Additional file 2 Author contributions C.X. and L.W. designed the study. W.Z., L.J., and Z.Y. did the bioinformatics analysis. H.J., L.Y., H.S., L.J., X.J., R.W., and J.W. collected patient information and serum. W.Z and C.X. written the article. All authors read and approved the final version of the manuscript. Funding This study is supported by the Medical Scientific Research Foundation of Guangdong Province (No.B2025492); Guangdong Science and Technology Plan Project, Construction of High-level Biosafety Laboratories (No. 2021B1212030003); Guangdong Provincial Clinical Research Center for Tuberculosis (No. 2020B1111170014); Guangdong Science and Technology Plan Project, Construction of High-level Biosafety Laboratories (No. 2021B1212030010); Youth Innovative Talents Project in Colleges and Universities in Guangdong Province (No. 2024KQNCX209); and the Special Fund for Science and Technology Innovation Strategy of Guangdong province (No. pdjh2024b632). Data availability The RNA—seq data has been submitted to the Sequence Read Archive with the bioProject accession number PRJNA1229112. Declarations Ethics approval and consent to participate Ethical approval was granted by the Ethics Committee of the Center for Tuberculosis Control of Guangdong Province. Plasma specimens were collected and stored at Center for Tuberculosis Control of Guangdong Province after obtaining informed consent from the patients. The study followed the ethical principles of the Declaration of Helsinki. Competing interests The authors declare no competing interests. Footnotes Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Zhuhua Wu, Jing Liang and Yuchuan Zhao contributed equally. Contributor Information Weilong Liu, Email: weilong8380@hotmail.com. Xunxun Chen, Email: grace_chen514@163.com. References Associated Data Supplementary Materials Additional file 1 Additional file 2 Data Availability Statement The RNA—seq data has been submitted to the Sequence Read Archive with the bioProject accession number PRJNA1229112.