Introduction
Endobronchial ultrasound is a minimally invasive technique to obtain tissue via ultrasonic bronchoscope to visualise and perform a real-time transbronchial needle aspiration under direct endobronchial ultrasonographic guidance [1 ]
[2 ].
Endobronchial ultrasonography with transbronchial needle aspiration is an established tool in mediastinal staging [3 ]
[4 ]
[5 ]. Apart from diagnosis of lung cancer, EBUS is appealing to physicians interested in diagnosis of other pathologies in the mediastinum, including malignancies despite lung cancer or non-malignant diseases like lymph node tuberculosis or sarcoidosis [6 ]
[7 ]
[8 ]
[9 ].
Compared to invasive surgical techniques such as mediastinoscopy, EBUS-TBNA appears to be safer and less expensive, with high specificity for identifying mediastinal lung cancer [10 ].
Solitary use of imaging methods in cancer patients, like conventional contrast enhanced computed tomography (CT) or positron emission tomography (PET) lacks accuracy and histopathological proof [11 ]. Originally, EBUS-TBNA was performed with a 22 gauge (22G) needle. The benefit from larger units such as the 21G needle or forceps has been discussed [12 ]. The impact of EBUS-TBNA as an integrated tool in the observation of non-malignant lymphadenopathies is rising, but less apparent than in diagnosis of malignant diseases.
In stage I/II sarcoidosis, EBUS-TBNA was associated with a sensitivity and accuracy of approximately 80 % [13 ].
EBUS-TBNA demonstrated high specificity of 100 % and accuracy of 91 % and was recommended as a safe and well tolerated procedure in tuberculosis [9 ].
In lymphoma patients, EBUS-TBNA may differentiate lymphoma relapse in the mediastinum from alternative pathologies and may therefore prevent surgical biopsies in most such patients [14 ].
As a further diagnostic advantage besides the cytopathological evaluation of the obtained specimens by EBUS-TBNA, additional microbiological testing was implemented in specific cases in this study. In previous studies, a modification of sampling techniques was recommended and more sensitive detection methods may be required for routine microbiological tests on EBUS-TBNA to gain a sufficient level of sensitivity and to rule out infectious causes of lymphadenopathy [15 ].
The aim of this study was to ascertain the diagnostic yield of EBUS-TBNA with cytopathological and microbiological testing in the diagnosis of non-malignant lymphadenopathy. Secondary endpoints included quality and safety of the procedure.
Material and Methods
Study design
A retrospective, single-centre, observation analysis of all endoscopic ultrasound bronchoscopy procedures was performed in a university medical centre between 12/03/2013 and 29/07/2015. Primary endpoint was the sensitivity of EBUS diagnosis on the basis of the final diagnosis 6 months after the procedure. Secondary endpoints included risk factors related to the development of complications.
The study adhered to ethical guidelines of the 1975 Declaration of Helsinki and was performed with the approval of the Ethics Committee of Hannover Medical School (Number: 2565 – 2015).
Patients
All patients referred to the bronchoscopy unit of the Department of Respiratory Medicine for EBUS-TBNA between 12/03/2013 – 29/07/2015 were included.
All patients underwent computed tomography (some of them as PET fusion CT) generally a maximum of 3 months prior to EBUS-TBNA.
Non-malignant indications for EBUS-TBNA were retrospectively divided from suspected malignant disease by the EBUS performing pulmonologists.
Suspected non-malignant mediastinal lymphadenopathy was defined by lack of a known or suspected malignancy. Patients were categorized in the non-malignancy group when they had exclusively mediastinal lymphadenopathy over 5 mm in diameter or PET-positive lymph nodes. Presence of lung parenchymal mass, primary tumour signs or history of or current malignant disease were exclusion criteria. Malignancy was suspected if patients with mediastinal lymphadenopathy and PET-positive lymphnodes had any history of malignancies or tumour signs on pulmonary imaging based on the Fleischner guidelines for solid and subsolid nodules [16 ]
[17 ].
Procedure
Every EBUS procedure was performed in conscious sedation via the nasal or oral route by experienced pulmonologists. One bronchoscopist performed the procedure, while another performed needle puncture and specimen preparation.
During the study period, four pulmonologists performed EBUS-TBNA (TF, JG, MG, HS). The EBUS procedure was performed using a convex probe ultrasound bronchoscope, designed for lymph node staging with a 2.2 mm instrumental channel (Olympus® BF-UC180F). All lymph node needle aspirations were performed with single use 22G needle (NA-201SX-4022, Olympus® ).
Prior to procedure, lidocaine was administered for anaesthesia of upper airway and minimizing cough. Midazolam and propofol were given intravenously for adequate sedation. Pulse, blood pressure, respiratory rate and pulse oxymetric saturation were monitored.
A flexible videoscope (Olympus, Tokyo, Japan; Type P180, Q180, T180) was inserted and topical lidocaine was administered to trachea and bronchus. Areas with suspicious mucosal changes were biopsied on conventional bronchoscopy and material sent to pathology. Lower respiratory tract samples were collected and sent to microbiology and/or cytopathology.
After initial exploration of the bronchial tree by conventional video bronchoscopy, EBUS was performed to visualise suspicious lymph nodes/pulmonary mass previously identified in computed tomographic or positron emission tomographic images. Lymph node stations were described by region, the lymph nodes were classified by size and appearance and documented [18 ]. All lymph node stations accessible by EBUS were screened (anterior and superior mediastinum: stations 1, 2, 3, 4, and station 7; intrapulmonary: stations 10, 11, and 12) [19 ].
After identification of suspicious lymph nodes/pulmonary mass (short axis diameter > 5 mm positive on thoracic CT scan or PET) on endobronchial ultrasound pictures, the transbronchial needle was inserted under direct EBUS guidance. On average, we performed 3 needle passages per lymph node/mass with 15 – 20 needle excursions according to recent recommendations [20 ]. Sampling was performed using the dry suction technique, applying negative pressure suction on the proximal end of the needle after the stylet is removed with a pre-vacuum syringe. The aspirate was collected and smeared onto glass. The smear was alcohol-fixed and underwent Pappenheim staining and, depending on medical questions, also other stains (Periodic-acid-Schiff (PAS), Ziehl-Neelsen etc). No rapid-onsite cytology evaluation (ROSE) was performed.
Residual materials were preserved for subsequent cell block preparation, using the Gauting protocol [21 ]. Additional material was collected in normal saline for microbiological testing where clinically indicated.
Definition of diagnosis
Based on cytopathological findings of EBUS-TBNA aspirates, clinical diagnoses on hospital discharge (including results of bronchoalveolar lavage (BAL) and biopsy on conventional bronchoscopy) and six months after the procedure, the following diagnoses were made: bronchial carcinoma (tumour cells with pulmonary origin, with or without other signs of malignancy, such as necrosis and inflammation), other malignant disease (detected tumour cells with origin other than pulmonary, with or without other signs of malignancy, such as necrosis and inflammation), sarcoidosis (epithelioid non-caseating granulomas without necrosis) and mycobacterial infection (evidence of acid-fast bacterial infection and epithelioid granulomas, with or without necrosis or solely detected necrosis). Results were considered as reactive lymphadenopathy when inflammatory cells, such as lymphocytes or granulocytes, where detected, without presence of malignant cells or granulomas.
Absence of lymphocytes defines an inadequate sample.
Material was considered adequate if lymphocytes were detected or if definite diagnosis was achieved.
Statistical Analysis
Statistical analysis was performed using IBM SPSS Statistics version 22. Data were presented as medians (with 25th and 75th percentiles). Differences between groups were compared univariately using the Mann-Whitney U test, Fisher’s exact test or Pearsons’s Chi-square test as appropriate. Pathological and microbiological results were reviewed and compared to final diagnosis based on clinical/radiological follow-up. A positive cytological result was usually accepted as evidence, and the patients were treated accordingly. The sensitivity, specificity, and accuracy were calculated using the standard definitions. Statistically significant differences were determined using Chi2 for categorical variables and the t-test for independent samples for metric variables. A p-value of < 0.05 or less was considered as significant.
Results
Patient characteristics
Within the study period, 315 patients underwent 333 endobronchial ultrasound-guided fine-needle
aspirations (191 men, 124 women; median age 63 years, IQR (interquartile range) 54 – 71, range
10 to 84 years). ( [Table 1 ]). EBUS-TBNA was performed in 111 patients with suspected non-malignant outcome, and in 204 patients with malignant diseases (152 with suspected lung cancer and in 52 with suspected malignant disease other than lung cancer). Malignant results were included to preserve comparability and quality assurance.
Table 1
Patient characteristics.
Indication
total
“non-malignant disease”
“malignant disease”
p-value
315 (100)
111 (100)
204 (100)
Gender, n (%)
Male
191 (61)
57 (51)
134 (66)
0.013
Age, median (IQR)
63 (54 – 71)
56 (45 – 65)
66 (56 – 72)
0.04
Smoking, n (%)
99 (31)
21 (19)
78 (38)
< 0.001
Long-term oxygen therapy, n (%)
7 (2)
2 (2)
5 (2)
0.709
Comorbidities, n (%)
COPD
74 (23)
17 (14)
57 (29)
0.003
Congestive heart failure
18 (6)
8 (7)
10 (5)
0.4
Lung transplantation
15 (5)
13 (12)
2 (1)
< 0.001
Sarcoidosis
3 (1)
3 (3)
0 (0)
0.018
Pulmonary hypertension (PAPm ≥ 25 mmHg)
15 (5)
11 (10)
4 (2)
0.002
Interstitial lung disease
3 (1)
3 (3)
0 (0)
0.018
Data are shown as no. (%) or median (interquartile range).
In the suspected non-malignant group, a lower percentage (19 %, n = 21) of patients were current smokers and, in contrast to patients with suspected malignant disease, generally younger. Frequent comorbidities are shown in [Table 1 ], in case of sarcoidosis as comorbidity EBUS-TBNA was performed for conformational purposes, when previous diagnosis was based on other methods or for exclusion of another disease.
Details of procedure and lymph nodes are shown in [Table 2 ] and [Fig. 1 ]. Preferred lymph node stations of a total of 245 lymph node biopsies in 111 patients were station 7, n = 93 (38 %), 11 L (left), n = 55 (22 %) and 4 R (right), n = 32 (13 %).
Table 2
Intervention, lymph node characteristics and safety.
Variables
total
“non-malignant”
“malignant”
P-value
(333)
(115)
(218)
Bronchoscopy duration (min), median (IQR)
30 (25 – 38)
28 (21 – 36)
32 (26 – 39)
0.556
Sedation
Midazolam alone, n (%)
32 (10)
16 (14)
16 (7)
0.055
Propofol, n (%)
295 (89)
96 (83)
199 (91)
0.018
Mean Midazolam dose, i. v. (mg)
3 (3 – 4)
3 (3 – 4)
3 (3 – 4)
0.029
Mean Propofol dose, i. v. (mg)
180 (120 – 250)
180 (135 – 255)
170 (120 – 243)
0.998
Diameter of largest LN (mm), median (IQR)
15 (10 – 20)
15 (10 – 19)
15 (11 – 20)
0.07
Number of examined LN per patient, median (IQR)
2 (1 – 2)
2 (1 – 2)
2 (1 – 2)
0.809
Number of LN-biopsies per patient, median (IQR)
5 (4 – 7)
5 (4 – 6)
5 (4 – 7)
0.108
Number of LN-biopsies per LN, median (IQR)
3 (2 – 4)
3 (2 – 4)
3 (2 – 3)
0.281
Adverse events, n (%)
36 (11)
9 (8)
27 (12)
0.297
Significant bleeding ( ≥ 50 ml)
22 (7)
4 (3)
18 (8)
0.095
Oxygen desaturation < 90 %
13 (4)
5 (4)
8 (4)
0.761
Cardiac arrhythmia
1 (0)
0 (0)
1 (0)
0.467
Data are shown as No. (%) or median (interquartile range). LN: lymph node
Fig. 1 Clinical diagnosis on hospital discharge and final diagnosis after 6-month follow-up. Data shown as no. (%).
Adverse events
Significant bleeding (≥ 50 ml) (n = 4, 3 %) and oxygen desaturation (< 90 %) (n = 5, 4 %) were reported as adverse events (AE). All AEs could be managed in the bronchoscopy suite.
Microbiological results
Microbiological testing was performed when occurred infection was considered possible (signs of inflammation, increased inflammatory values in the blood) and was conducted on 73 lymph nodes, resulting in 52 positive cultures. The most common pathogens found were Streptococcus species (sp.) and Staphylococcus sp. (n = 56), followed by Veillonella sp. (n = 9).
Relevant infections (Mycobacteria tuberculosis, Nocardia nova), as defined by according treatment following the procedure, were found in three cases. Other organisms and non-respiratory pathogens were considered as contaminants.
Microbiological testing of bronchoalveolar lavage and collected bronchial secretions was performed in 97 patients and revealed positive bacterial colonisation in 52 cases. In 9 out of 52 cases, lower respiratory tract samples and lymph node samples revealed an identical pathogen.
Diagnostic yield
The clinical diagnosis on hospital discharge and the final diagnosis after 6 months of
follow-up are shown in [Table 3 ].
Table 3
Results.
Indication
total
"no malignancy“
"malignancy“
Clinical diagnosis on hospital discharge
333
115 (35)
218 (65)
Lung cancer
124 (37)
8 (7)
116 (53)
Other malignant disease
22 (7)
2 (2)
20 (9)
Sarcoidosis
23 (7)
18 (16)
5 (2)
Mycobacteria
8 (2)
5 (4)
3 (1)
Pulmonary cyst
3 (1)
1 (1)
2 (1)
Nocardia nova
1 (0)
1 (1)
0 (0)
Reactive Lymphadenopathy
134 (40)
73 (63)
61 (28)
Inadequate sample
18 (5)
7 (6)
11 (5)
Final diagnosis follow-up 6 months
Lung cancer
141 (42)
9 (8)
132 (61)
Other malignant disease
35 (11)
3 (3)
32 (15)
Sarcoidosis
30 (9)
25 (22)
5 (2)
Mycobacteria
13 (4)
7 (6)
6 (3)
Benign/pulmonary cyst
3 (1)
1 (1)
2 (1)
Nocardiosis
1 (0)
1 (1)
0 (0)
Non-specific inflammation
85 (26)
54 (47)
31 (14)
Loss to follow-up
25 (8)
15 (13)
10 (5)
Data are shown as no. (%).
The calculated sensitivity, specificity, accuracy, positive predictive value (PPV) and negative predictive value (NPV) after EBUS examination in the “non-malignant” group (n = 115) were 76 %, 96 %, 87 %, 95 % and 83 %.
18 cases had sarcoidosis, ten cases had malignant diagnosis, and five had mycobacterial
infection. In addition, there was one cyst formation and one case of Nocardia infection.
([Fig.1 ])
In our study, sensitivity of sarcoidosis was 72 % in the “non-malignant” group and accuracy 93 %.
In our study, unfortunately, in 7 patients with a later diagnosis of sarcoidosis during follow-up, the EBUS-TBNA result was considered as reactive lymph nodal enlargement (cytopathological testing showed lymphocytes). 3 of the patients received additional (definite) diagnosis by surgical lymph node sampling or CT-guided lymph node sampling. In the other 4 cases, diagnosis of sarcoidosis was made by other methods.
In five out of seven patients with tuberculosis, diagnosis could be made by EBUS-TBNA (sensitivity = 71 %, accuracy = 98 %). In 3 patients cytopathological testing was positive for mycobacterium (epithelioid granulomatous reaction and caseous necrosis), microbiological testing of lymph node was negative. In two out of three cytopathological-positive patients microbiological testing of BAL showed positive results for Mycobacterium tuberculosis. In the remaining cytopathological-positive patient tuberculosis infection had previously been reported. 2 cases were diagnosed by microbiological detection of acid-fast Mycobacterium tuberculosis in the lymph nodes. In 2 cases, EBUS-TBNA was inconclusive and failed to make the final diagnosis of tuberculosis.
The single case in our study with a positive microbiological result of Nocardia nova in the lymph node examination had pulmonary infiltrates and inflammatory signs in the cytopathological diagnosis of lymph nodes. BAL was performed, but culture remained negative. The patient was suffering from iatrogenic Cushingʼs syndrome, which resulted in steroid-induced diabetes mellitus. Both conditions could be predisposing factors for Nocardia infection [22 ].
There was one cyst formation finally diagnosed by thoracotomy, in which cytopathological examination of the EBUS-TBNA aspirate showed lack of lymphocytes with suspicion of a cyst and inflammatory cells (lymphocytes, granulocytes and alveolar macrophages).
Sensitivity of malignancy in the “non-malignant” group was 83 %; in two out of twelve cases, EBUS-TBNA did not result in the diagnosis of cancer.
Discussion
EBUS-TBNA is a useful tool in the diagnostic work-up of lymphadenopathy in general and besides suspicion of cancer disease. Apart from histopathological analysis of specimens, additional microbiological testing of EBUS-TBNA should be performed in the work-up of suspected non-malignant diseases.
With a sensitivity of 76 % and specificity of 96 % in patients with suspected non-malignant conditions and no severe adverse events present, EBUS-TBNA appears to be a safe and accurate method for the evaluation of cytopathological changes in lymph nodes and the identification of lymph node microbiological infection.
Non-diagnostic results should be controlled by follow-up.
In the diagnosis and staging of lung cancer, EBUS-TBNA is a well-established tool in contrast to its value in non-malignant diseases [23 ]. Recent publications with a focus on diagnostics of malignancy reported a similar sensitivity but higher specificity (sensitivity and specificity ranging from 80 – 95 % and specificities 98 – 100 %) compared to our results [24 ]
[25 ]
[26 ]
[27 ]. For quality assurance reasons, data from suspected malignant disease was included in our study. The “malignant” group’s sensitivity was 82 % and specificity 97 %, showing a higher sensitivity and specificity compared to the “non-malignant” group. Another recent study describes higher sensitivity in the diagnosis of malignant diseases than benign diseases (92.6 % vs. 89.7 %) [4 ].
In sarcoidosis, EBUS-TBNA in mediastinal lymphadenopathy is established and studies recommend EBUS-TBNA in combination with standard bronchoscopy as a first-line examination [28 ]
[29 ]. In our study, sensitivity of sarcoidosis was 72 % in “non-malignant” group and accuracy 93 %. 17/25 patients were diagnosed by EBUS-TBNA alone (sensitivity of 68 %), there was one additional case of sarcoidosis detected by endobronchial biopsy (EBB). A meta-analysis by Agarwal et al. showed a pooled sensitivity of 78 % [30 ]. A similar result is seen in a granuloma trial, which confirmed EBUS-TBNA with a sensitivity of 74 % to be preferred over conventional biopsies in the diagnosis of granulomas [31 ].
Caglayan et al. stated that sensitivity depends on the experience of the operator, stage of sarcoidosis (higher in Stage II than in Stage I) and the number of lymph nodes examined (single lymph node station: 72.7 % for single lymph node station to 85.3 % for examination of more than one lymph node). This study describes a higher overall sensitivity for granulomatous mediastinal lymphadenopathy (81 %) [32 ].
For tuberculosis, in our study sensitivity revealed 71 %, with an accuracy of 98 %.
In the diagnosis of tuberculosis, EBUS-TBNA is considered to be the method of choice, as a safe procedure with a high diagnostic yield [33 ]. Previous studies describe a composite microbiological and clinicopathological diagnosis of 69 % to 86 % in TB-endemic countries [7 ]
[34 ].
There was one Nocardia nova infection, which was solely detected in lymph node aspirate and not in BAL/secretion. To the best of our knowledge, this is the first study describing Nocardia nova infection in lymph nodes in an immunocompromised patient. Nocardia infections are usually found as an opportunistic infection in immunocompromised patients, but in one third it can occur in immunocompetent patients. When isolated from the respiratory tract, it should not be regarded as a contaminant. Occasionally it may be identified from the respiratory tract without apparent pulmonary infection [35 ].
Relevance of Nocardia nova isolated in mediastinal lymph nodes remains unspecified.
Recent publications described the utilisation of EBUS-TBNA as a safe and accurate tool even in immunocompromised conditions for the identification of associated diseases such as pulmonary mycoses/Cryptococcosus, as well as Nocardia infection (N. asteroides, N. beijingensis and N. arthritidis) [36 ]
[37 ]
[38 ].
One case of mediastinal cyst formation, already suspected by EBUS-TBNA and confirmed by thoracotomy, was present in the “non-malignant” group. Mostly cysts are asymptomatic, but they can also be infectious and show symptoms.
For exclusion of malignancy, surgical methods are often recommended in the treatment of mediastinal cyst formation, although related with higher complication risks. However, via EBUS, it is possible to drain the cysts more safely, to explore the diagnosis of mediastinal structure [39 ]. Other publications state that drainage of cysts by EBUS-TBNA is related to a higher risk of infectious mediastinitis compared to the aspiration of lymph nodes [40 ]
[41 ].
In the case of microbiological testing, our findings showed relevant pathogens solely in the lymph nodes in two out of three cases, not being detected in patient’s secretions (gathered by bronchial lavage), may point to the importance of routine microbiological testing.
Despite this fact, due to frequent bacterial contamination and inadequate sensitivity, other publications suggest that routine microbiological testing is being reserved for populations with higher probability for Mycobacterium tuberculosis, Mycobacterium avium infection or endemic fungal infections [15 ].
In the “malignant” group, microbiological evaluation of biopsies and lower respiratory tract samples showed acid-fast bacilli in three cases. Two cases of non-tuberculous infections (M. Kansasii, M. avium) were detected, in addition to one Mycobacterium tuberculosis infection.
Limitations
Several limitations of our study should be noted.
This is a retrospective, single institution, non-randomised study, with a previous selection of patients into two groups (malignant vs non-malignant). The group-classification of each patient was made by the EBUS team of four experienced pulmonologists retrospectively. With this subjective assigning selection biases are possible to arise.
The decision to undertake EBUS-TBNA was at the discretion of the referring physician; the referral may vary or be related to patient preferences introducing selection biases as well.
There was no pre-selection of suitable cases, and examinations were conducted in routine practice.
The study size was not formally calculated, but is greater than other comparable studies [42 ]
[43 ].
All patients referred consecutively for investigation of mediastinal lymphadenopathy were included in the study and only 6 % of examinations were not representative. Depending on examination conditions, the operator’s experience and artefactual changes (for example due to transport time), the quality of examination result differs [44 ]
[45 ].
The follow-up period was defined to be 6 months after procedure and based on clinical/radiological diagnosis. In our retrospective study we count 25 cases in total which are lost during follow-up, most cases (n = 15) in the “non-malignant” group, which could be explained by non-necessity of Re-EBUS-TBNA or other diagnostic method in the case of negative EBUS-TBNA results.
Persistence of diagnosis during this period of time is not always ensured due to the fact that benign lymphadenopathy shows progression at times. For example, can a sarcoidosis become inactive, reactive lymphadenopathy can be a precursor of a developing disease or disappear without treatment. Significant differences between clinical diagnosis on hospital discharge and follow-up were seen in the diagnosis of sarcoidosis and tuberculosis. Due to the histopathological heterogeneity of non-malignant diseases, compared to typical malignant pathological findings, often further diagnosis is needed to make the final diagnosis.
Outlook
As a minimally invasive technique, only a small sample size can be collected by EBUS-TBNA. In our study, preparation of cell blocks was performed to improve the quality of diagnostic material of remaining cytology. It showed a slightly larger number of representative results (86 % compared to 85 % of smear results).
Collection of larger tissue samples might be relevant to overcome low sample adequacy. Studies have reported a better diagnostic yield by using mini-forceps rather than a 21G needle, and no difference in adequacy between 21G and 22G needles [46 ]
[47 ]. Statistical superiority of mini-forceps compared to conventional TBNA by 22G needle have also been reported by Herth et al. in diagnosis of lymphoma and sarcoidosis [48 ].
A study of Shiu et al. stated a higher diagnostic yield of a 19G needle compared to a 22G needle, at least in the examination of lymph nodes in non-cancer patients [49 ].
These results highlight the relevance of larger tissue size in order to improve diagnostic outcome in non-malignant diseases.
This study demonstrates actual diagnosis in a population in which malignancy is not previously expected and against expectations it shows a significant number of malignant results in the end. This assumes EBUS-TBNA to be a useful tool in diagnosing lymphadenopathy other than from suspected underlying cancer disease. The calculated low NPV shows that patients with EBUS-negative lymphadenopathy need careful follow-up. Particularly in the case of sarcoidosis and tuberculosis, confirmation by other methods might be indicated. Its role in diagnosing infections has to be further evaluated.
