Endobronchial Ultrasonography Features in Tuberculous Mediastinal Lymphadenopathy
Article information
Abstract
Background
Mediastinal lymphadenopathy indicates an intrathoracic pathological process. Tuberculous mediastinal lymphadenopathy (TML) can be an early sign of tuberculosis infection, particularly in developing countries. Its incidence is often unclear due to underdiagnosis and limited access to diagnostic tools. Timely diagnosis is essential, as delayed treatment can result in severe complications. Endobronchial ultrasound-guided transbronchial needle aspiration (EBUS-TBNA) is a reliable method for obtaining samples and establishing a diagnosis, although its use is constrained by various factors. This study aims to clarify the role of EBUS features in assessing TML.
Methods
This is a cross-sectional study that recruited ≥18 years old patients with suspected TML who underwent EBUS-TBNA. Mediastinal lymph nodes were systematically evaluated EBUS, with a focus on the largest nodes for further characteristic analysis. A TBNA was performed, and the retracted samples were assessed for pathology, acid-fast bacilli, Mycobacterium tuberculosis (MTB) culture, and Xpert Ultra examinations.
Results
One hundred patients were recruited for the study. Fifty-three of these patients were ultimately diagnosed with TML. Significant associations were found between TML and several factors: short-axis size (p<0.001), oval-shaped lymph nodes (p=0.034), indistinct margins (p<0.001), central hilar presentation (p<0.001), internal echoes (p<0.049), thin layer echogenicity (p=0.033), and nodal matting (p<0.001). Among the diagnostic modalities, Xpert Ultra demonstrated the highest sensitivity for TML at 71.7%.
Conclusion
Endobronchial ultrasonographic features such as small size, round shape, indistinct margin, central hilar presentation, internal echo, thin layer echogenicity, and nodal matting are indicative of TML.
Introduction
The lymph node is an immunological organ that plays a role in the body’s defense and can reflect abnormalities in the organs or tissues it supplies [1]. In response to antigens, the body will mount a proliferative response, resulting in the formation of a larger structure known as lymphadenopathy [2]. Mediastinal lymphadenopathy may be found incidentally and requires further evaluation, as it can be an early sign of a thoracic disorder. Its prevalence varies depending on the underlying etiology, whether benign or malignant [3]. Tuberculosis (TB) remains a significant global health burden. While involvement of mediastinal lymph nodes is typically a complication of TB, it can also manifest as isolated mediastinal lymphadenopathy without lung involvement [4,5]. Tuberculous lymphadenopathy is the most common form of extrapulmonary TB, accounting for 30% to 40% of total TB cases. However, the proportion of tuberculous mediastinal lymphadenopathy (TML) remains unknown [6]. Due to its rarity, the diagnosis of TML is sometimes overlooked [7].
Diagnostic evaluation is often challenging and requires complex modalities [8]. Mediastinoscopy was once the gold standard diagnostic procedure; however, due to its invasive nature, it is rarely performed today. Conversely, less expensive procedures, such as conventional transbronchial needle aspiration (cTBNA), also carry some risk of complications [9,10].
However, if it is not diagnosed immediately, TML can potentially lead to fatality by eroding surrounding structures, such as the pleura, vertebrae, esophagus, pericardium, and heart, which may result in respiratory failure [11-13]. Endobronchial ultrasound (EBUS) is a minimally invasive procedure that uses ultrasound imaging to visualize the airway walls and surrounding structures, such as mediastinal lymph nodes. This technique is associated with a higher diagnostic yield compared to cTBNA [14,15]. EBUS is typically performed in conjunction with TBNA procedures (EBUS-TBNA) to enhance the quality of the results [16]. However, in fact, the utilization of EBUS-TBNA occurs in less than 50% of patients with mediastinal lymphadenopathy [17]. The low percentage is due to several factors, including cost issues, operator skills, technical difficulties, small lymph node size, inadequate sample predictions, and laboratory examination capabilities [17]. Therefore, the characteristics of endobronchial ultrasonograms of mediastinal lymph nodes are expected to aid in predicting diagnostic and treatment decisions, in TML.
Ultrasonographic features of mediastinal lymph nodes can be valuable for disease prediction in clinical practice. This is particularly relevant in situations where performing EBUS-TBNA is not feasible due to the unavailability of the EBUS tool, lack of a skilled operator, limited resources, inadequate or inconclusive results, or when it is impractical to repeat the EBUS-TBNA procedure or to conduct further invasive procedures like mediastinoscopy. If the initial biopsy result is inconclusive, the ultrasonographic characteristics of the mediastinal lymph node may help determine whether a repeat biopsy is necessary [18]. This study aims to evaluate the role of endobronchial ultrasonographic features in assessing TML.
Materials and Methods
1. Study design and patients
A cross-sectional study was conducted at a national respiratory referral hospital from July to December 2024. The EBUS-TBNA procedure was performed as a standard diagnostic approach for lung and mediastinal diseases. Institutional ethics committee approval was obtained for this study. Patients who met the inclusion criteria were recruited using a non-probability sampling technique.
Patients who were suspected of having TML were adult patients (≥18 years old) who have mediastinal lymphadenopathy (≥5 mm short-axis diameter) with one or more of the following criteria:
(1) Clinical symptoms suggestive of TB include respiratory symptoms (cough, dyspnea, or chest pain) and systemic symptoms (fever, loss of appetite, or weight loss).
(2) Persistent mediastinal lymphadenopathy following previous non-TB treatment.
(3) Patients with typical TB lung lesions who had been prescribed anti-TB drugs by their previous respirologist and had been taking them for less than 60 days.
(4) Asymptomatic with one risk factor for TB (malnutrition, diabetes mellitus, close contact with TB, or smoking history).
2. Endobronchial ultrasound
Preparation and procedures for EBUS were based on the standard operating procedures of the institutional hospital. The procedure was performed by a single pulmonologist to minimize bias. General anesthesia or moderate sedation was used, depending on the patient’s condition. The EBUS machine utilized was the Fujifilm SU-1-S Endoscope Ultrasonic Processor (frequency 7.5 MHz; Fujifilm, Tokyo, Japan) along with the Fujifilm EB530ES linear type EBUS scope. After inserting the EBUS scope into the trachea, the balloon attached to the transducer was inflated to ensure full contact with the mucosa and improve visualization.
Mediastinal lymph node stations were evaluated systematically with the order of:
(1) Station 2R (right upper paratracheal)
(2) Station 4R (right lower paratracheal)
(3) Station 7 (subcarinal)
(4) Station 4L (left lower paratracheal)
(5) Station 2L (left upper paratracheal).
Lymph nodes at all stations were measured. If multiple lymphadenopathies were present, the longest one was analyzed further and noted using a standardized checklist table. This checklist consisted of 12 ultrasonographic features that needed to be analyzed in real time: shape, size, border, central hilus, echoic texture, posterior acoustic features, echogenicity, internal echo, thin echogenic layer, nodal matting, perinodal edema, and vascularization. Further details on each feature are provided in Supplementary Table S1. The procedure was documented using recording systems connected to the processor, and the video images were stored in digital imaging and communications in medicine (DICOM) format. The recorded images were then evaluated in a non-real-time manner by another pulmonologist. If there was a difference in judgment regarding certain items between the two pulmonologists, those items were further evaluated by a third pulmonologist to determine the final decision based on either the first or second operator’s assessment.
3. Endobronchial ultrasound-guided transbronchial needle aspiration
After evaluating the ultrasonogram features, EBUS-TBNA was performed using a 21G EBUS-TBNA (Vizishot; Olympus needle, product code NA-201SX-4022, Center Valley, PA, USA). Once the targeted lymph node was visualized, the EBUS-TBNA needle was inserted through the working channel of the EBUS scope until the sheath was visible. The needle was then advanced into the lymph node to a depth adjusted according to the targeted hypoechoic lesion, which indicated tissue necrosis. If the structure was not identified, the needle was directed to the center of the node. Next, the stylet was removed from the needle, and a 20-cc syringe was attached as a suction system while performing a minimum of 30 agitations. Agitation was carried out using a fanning technique, repeatedly directing the lever up and down to obtain a larger sample. Each lymph node was penetrated by the TBNA needle at least three times until an adequate sample was obtained, defined as multiple filaments measuring 2 cm in length based on macroscopic assessment [19,20]. The collected specimens were expelled from the needle into a 10% formalin fixation solution using a stylet and syringe, and then processed into cell blocks for microscopic assessment. They were immediately transferred to the anatomical pathology laboratory for paraffin block preparation and hematoxylin-eosin staining. Any residual specimen in the needle catheter was rinsed with saline solution for TB panel examinations, including acid-fast bacilli (AFB), Mycobacterium tuberculosis (MTB) Mycobacteria Growth Indicator Tube (MGIT) culture, and Xpert Ultra. This was promptly delivered to the microbiology laboratory. If any complications arose, such as uncontrolled bleeding or hemodynamic instability, the procedure would be halted.
4. Laboratory examination
Histopathological diagnostics were performed by a single pathologist. A sample was considered microscopically adequate if it contained more than 40 lymphocytes per 10 large fields of view (magnification ×40), more than 100 lymphocytes per 5 small fields of view (magnification ×100), or if it included central germinal fragments [18,20]. If none of those features were found, the samples were still considered adequate if a clear diagnostic feature, such as granuloma or malignancy, was identified [21,22].
For the microbiology test, saline water containing residual specimens was rinsed and placed into a sterile pot. This procedure can be repeated until a 5 mL sample is collected. AFB can be visualized under a microscope using carbolfuchsin staining via the Ziehl- Neelsen method. After the decontamination procedure and re-suspension with phosphate buffer, some samples will be inoculated into 0.5 mL MGIT tubes. The Xpert MTB/RIF Ultra test was performed using the GeneXpert 6-color device (Cepheid, Sunnyvale, CA, USA).
5. Determination of final diagnosis
The final diagnosis was based on clinical, histopathological, and microbiological records. The referring physician will be contacted to provide clinical information that would assist in the patient’s management. A patient was considered to have a diagnosis of TML if they met one or more of the following criteria:
(1) A positive AFB or MGIT culture test.
(2) A positive Xpert MTB/RIF Ultra test.
(3) Clinical symptoms, risk factors, and histological changes (granulomatous inflammation) were highly suspicious for TB, warranting anti-TB treatment.
6. Statistical analysis
IBM SPSS software product version 30.0 for Mac OS (IBM Co., Armonk, NY, USA) was used for analyzing the data. Qualitative data were presented by number and percentage, while qualitative data were presented by mean, median, standard deviation and interquartile range (based on normality of the data). Both parametric and non-parametric analysis were used for significance testing. Sensitivity and specificity were determined on a per-patient basis. A p-value <0.005 was considered significant.
7. Ethics approval
This study was approved by the Ethics Committee of Health Research at Persahabatan Hospital (Ethical Approval No. 0189/KEPK-RSUPP/07/2024) and was conducted in accordance with the principles of the Declaration of Helsinki. Written informed consent from the patients was waived due to the retrospective nature of our study.
Results
From July 2024 to December 2024, 188 EBUS-TBNA procedures were performed, of which 109 were suspected cases of TML. Six patients were excluded because they had already received anti-TB drugs for more than 60 days, and three patients were under 18 years old, resulting in a final cohort of 100 patients (Figure 1). The proportion of male to female patients was equal, with the highest age range being 18 to 30 years. Twenty-eight patients had pulmonary TB co-infection based on radiological results, and seven patients also had cervical TB lymphadenopathy. Thirty patients were underweight, and 44% were smokers. Almost all patients exhibited clinical symptoms, predominantly cough (83%), weight loss (64%), and fever (35%). Only five patients had no symptoms and were incidentally found to have mediastinal lymphadenopathy. Seventy-four patients had comorbidities, primarily malnutrition (32%) and type 2 diabetes mellitus (30%). Three patients had reactive human immunodeficiency virus (HIV) serology results. Imaging results indicated that mediastinal widening was the most frequently recorded finding, followed by lung nodules/consolidation and infiltrates.
Patient selection flowchart. EBUS-TBNA: endobronchial ultrasound-guided transbronchial needle aspiration; ATD: anti-tuberculosis drug.
Most patients (70%) had multiple mediastinal lymphadenopathies, with the longest short-axis diameters observed in the subcarinal lymph nodes (44%) and right lower paratracheal lymph nodes (42%). The mean short-axis diameter was 10.84 mm, ranging from 5 to 20 mm. The proportion of lymph node enlargement varied by location: 30% for station 2R, 4% for station 2L, 74% for station 4R, 33% for station 4L, and 71% for station 7. Most procedures (95%) were performed under general anesthesia, and 57% of patients exhibited no abnormalities in the airway mucosa. Nineteen patients had stenotic airways, 13 had cicatricial lesions, and 10 had purulent secretions. No procedural complications occurred. The majority of mediastinal lymphadenopathy features included oval shape (64%), size of 5 to 9 mm (49%), indistinct margins (62%), presence of central hilar structures (58%), isoechoic appearance (55%), posterior acoustic shadowing (64%), heterogeneous echogenicity (59%), internal echoes (75%), thin echogenic layers (68%), nodal matting (51%), perinodal edema (78%), and capsular vascularization patterns (41%).
Fifty-three patients (53%) received a final diagnosis of TML. Ten patients tested negative for microbiological diagnostics (AFB, MTB culture, or Xpert Ultra). Their diagnoses were supported by histological changes indicating granulomatous inflammation and a high suspicion of TB based on risk factors, clinical presentation, and radiological features (Table 1). Granulomatous inflammation was observed in 32 patients, 17 of whom exhibited features of caseous necrosis. The majority of pathological results were non-specific or indicative of reactive lymph nodes (36%). Fifteen patients had malignant cells identified in their lymph nodes. MTB was detected in four patients via Ziehl-Neelsen stain, 24 patients via MGIT MTB culture, and 38 patients via Xpert Ultra, all showing sensitivity to rifampicin.
Demographic and clinical features for suspected tuberculous mediastinal lymphadenopathy
Forty-seven patients (47%) were diagnosed with non-TML, including conditions such as metastatic lung cancer, metastatic lung non-cancer, reactive lymph nodes in bacterial pneumonia, lung aspergillosis, heart failure, lung candidiasis, and reactive lymph nodes of unknown cause. Further details are available in Supplementary Table S2.
Twelve ultrasonographic features were evaluated in this study. Seven features were significantly associated with TML: size <10 mm (p<0.001), round shape (p=0.033), indistinct margin (p<0.001), central hilar presence (p<0.001), internal echo (p=0.049), thin echogenic layer (p=0.033), and nodal matting (p<0.001). The median short-axis diameter was 8 mm (range, 5 to 20) in the TML group and 11 mm (range, 5 to 25) in the non-TML group. No specific characteristics related to TML were identified in terms of echoic structure, posterior acoustic shadowing, echogenicity, perinodal edema, or vascularity. Ultrasonographic characteristics are reported in Tables 2, 3, while the comparison of ultrasonographic features between TML and non-TML is reported in Table 4.
Tuberculous mediastinal lymphadenopathy with negative microbiologic diagnostic
Ultrasonographic features in suspected tuberculous mediastinal lymphadenopathy
Ultrasonographic features in TML and non-TML
Multivariate analysis revealed that the ultrasonographic features associated with TML included lymph node size (odds ratio [OR], 8.72; p<0.001) and nodal matting (OR, 7.15; p<0.001). Receiver operating characteristic (ROC) curve assessment indicated that the area under the curve (AUC) for the regression equation was 87.9% (95% confidence interval, 81.2% to 94.5%), indicating strong predictive value. Additionally, the diagnostic profiles for bacteriological and pathological tests were calculated, showing that Xpert Ultra had the highest sensitivity (71.69%), followed by granulomatous inflammation (60.37%), MGIT MTB culture (45.28%), and AFB (7.54%). Detailed diagnostic profiles are listed in Table 5.
Diagnostic profiles in tuberculous mediastinal lymphadenopathy
Discussion
The final diagnosis of TML was confirmed in 53 patients. Among these, 10 exhibited granulomatous inflammation and were clinically suspicious for TB without bacteriological confirmation, while the final diagnoses for the remaining 47 patients varied (Supplementary Table S2). This indicates that although MTB detection remains the gold standard for diagnosing TB, we must also consider other supporting clinical evidence. A previous study reported similar findings, highlighting an increase in the sensitivity of EBUS-TBNA in TML from 62% microbiologically to 92% when the diagnostic standard was changed to include a combination of pathology and clinical features [23]. Although pathological examination commonly shows higher sensitivity, the results do not provide true-positive confirmation of TB, as the findings can also be associated with other infections or diseases [24,25].
Almost all patients exhibited clinical symptoms, with only five patients being asymptomatic. This result differs from the study by Navani et al. [26], which reported that 22% of its patients were asymptomatic. This outcome was anticipated since the study was conducted at a national referral hospital, where referral of asymptomatic patients from primary care is limited by national insurance guidelines. Among the 12 ultrasonographic features evaluated, TML was significantly associated with seven characteristics: small size, round shape, indistinct margin, central hilar localization, internal echo, thin layer, and nodal matting. In the TML group, the median axis of the lymph node was found to be 8 mm. This finding aligns with Hylton et al. [17], which demonstrated that a size of less than 10 mm is associated with a non-malignant prediction. The round shape was also the same form found in tuberculous cervical lymphadenopathy [27]. Round structures formed as the short to long axis ratio increased (>1/2) due to tissue inflammation and abscess formation, altering the normal structure typically observed in healthy or reactive lymph nodes.
In a previous study, it was stated that indistinct margins were significantly associated with the prediction of non-malignant lesions [17,28,29]. In TB, the unclear capsule lines were attributed to edema and inflammation of the surrounding soft tissue caused by inflamed lymph nodes (adenitis). In contrast, the distinct margins observed in malignant nodes are believed to result from tumor infiltration and a reduction in fatty deposition, which increases the acoustic impedance difference between the nodes and their surrounding tissues [30]. Central hilar regions were also significantly correlated with TML. The central hilar area is where blood and lymphatic vessels drain into the lymph nodes. The abundance of collecting sinuses in this area creates acoustic interfaces that reflect a portion of the ultrasound, making the hilus appear echogenic. Granulomatous inflammation in TB, even without extensive necrotic tissue, can still preserve the hilar structure. However, in malignant or abscessed nodes, the medullary sinuses may be disrupted, leading to the loss of the hilum structure. Consequently, the disappearance of this structure can also occur in tuberculous lymphadenopathy [31,32]. Internal echoes are strongly formed from nodal calcification or hyalinosis of caseous necrosis in TB. A previous study stated that these features are highly specific to tuberculous lymphadenopathy. They can be found either solitary or multiple, as well as central or peripheral [33]. In TB, it may be found in groups in a few locations or gathered as a single hyperechoic zone with posterior acoustic shadowing [34].
The formation of a thin layer structure can be attributed to a decrease in acoustic impedance between the lymph nodes and the surrounding tissue, likely caused by edema. This phenomenon results in a distinct peripheral margin around the nodes, typically measuring 2–3 mm in width (Figure 2) [27]. Mostly, it was primarily observed during the early inflammatory phase of TB [35]. It also correlated with the encapsulated line of granulomatous tissue surrounding the caseous necrosis, which is characterized by the proliferation of epithelioid cells, giant cells, and lymphocytes [36]. In the late stage of inflammation, destruction occurs between the nodes, leading to nodal confluence or nodal matting [35].
Sonogram of lymph node exhibiting a thin echogenic layer (arrows).
From the seven ultrasonographic features, two were highly correlated with TML. In the multivariate analysis, the short-axis size had an OR of 8.72, and nodal matting had an OR of 7.15, demonstrating strong predictive power as indicated by the ROC curve analysis (AUC, 87.9%). This suggests a good ability to predict TML lesions. Regarding granulomatous inflammation, which was the final diagnosis criterion of this study, the majority of patients in the third criteria group showed positive clinical and radiological responses to anti-TB drugs, as detailed in Table 2. In our study, the histopathological diagnosis of granulomatous inflammation demonstrated lower sensitivity compared to the Xpert Ultra results (60.4% vs. 71.7%). This finding contrasts with a study by Park et al. [37], which reported higher sensitivity for cytological results compared to Xpert Ultra (66.7% vs. 50%). Other studies have also concluded that cytological results exhibit better diagnostic accuracy than microbiological results [9,38]. This discrepancy may be attributed to several factors, such as the application of Xpert Ultra and the smaller size of the lymph nodes. Xpert MTB/RIF Ultra is more sensitive than Xpert MTB/RIF because it processes a larger sample volume, allowing for optimal PCR performance and the ability to bind to both wild-type and mutant sequences [39]. Additionally, the use of Xpert for TB diagnosis has advantages in our setting, as it reduces the median time to diagnosis (2 to 4 days) compared to histopathology (7 to 10 days) and MTB culture (20 to 46 days), and it is not affected by the number of tissue samples. However, the small size of lymph nodes in the TML group (ranging from 5 to 9 mm) may pose challenges for obtaining tissue samples, as indicated by a lower proportion of macroscopically satisfactory samples in the TML group compared to the non-TML group (81% vs. 83%). Unfortunately, six patients had inadequate sample quality microscopically, which was determined by the low quantity of lymphocyte cellular areas. Although only macroscopic satisfactory EBUS-TBNA samples (Matias et al. [20]) were included in this study, there remained a risk of inadequate histopathologic samples. This could happen due to the small size of the lesion (approximately 5 mm) or the failure to obtain a sample of the lymphoid tissue [21,22]. Even the mediastinoscopy procedure carries the same risk of obtaining inadequate tissue [40]. A limitation of this study is that we did not conduct rapid on-site cytology examination (ROSE) to enhance diagnostic capability. Future research is needed with a larger sample size to assess the predictive score of TML based on ultrasonographic characteristics observed in EBUS.
In conclusion, ultrasonographic features of mediastinal lymph nodes play a significant role in determining the cause of mediastinal lymphadenopathy. Characteristics such as small size, round shape, indistinct margins, central hilar echo, internal echoes, thin layers, and nodal matting suggest TML. Among these, small lesions and nodal matting are the most indicative of TML.
Notes
Authors’ Contributions
Conceptualization: all authors. Methodology: all authors. Formal analysis: all authors. Data Curation: Desianti GA, Mansyur M. Validation: Desianti GA, Rasmin M, Lisnawati L, Burhan E, Mansyur M, Karuniawati A. Investigation: Desianti GA, Rasmin M, Mansyur M, Andarini SL. Writing - original draft preparation: Desianti GA, Rasmin M, Mansyur M. Writing - review and editing: Desianti GA, Rasmin M, Burhan E, Mansyur M. Approval of final manuscript: all authors.
Conflicts of Interest
No potential conflict of interest relevant to this article was reported.
Funding
No funding to declare.
Supplementary Material
Supplementary material can be found in the journal homepage (http://www.e-trd.org).
Sonogram characteristic features checklist.
Lymphadenopathy mediastinal non-tuberculosis (n=47).
