Aim To assess the diagnostic role of endobronchial ultrasound (EBUS) guided transbronchial fine needle aspiration (TBNA) cytology, in the investigation of mediastinal lymphadenopathy and masses, with emphasis on pathology, criteria for adequacy, and use of liquid based cytology (LBC).
Methods In January 2008, EBUS guided TBNA was introduced for the staging of lung cancers and for the investigation of unexplained mediastinal lymphadenopathy and masses. Initially, the material was processed conventionally. In May 2008, the laboratory procured the Cytyc T2000.
Results 250 specimens (229 patients) were included from January 2008 to August 2009. The overall inadequate rate was 12%; 16.6% with the conventional method, 17.2% when both conventional and LBC were used, and 9.8% with LBC. With the conventional method, an average of 7.5 slides and one cell block were received per case. With LBC, one slide and one cell block were used for diagnosis. In 72 cases (28.8%) the aspirate showed no evidence of malignancy. In 122 cases (48.8%), there was metastatic disease. In five cases (2%), a diagnosis of atypia, suspicious for non-small cell malignancy, was made. One case (0.4%) each of amyloidosis and mediastinal goitre, 3 cases (1.2%) of lymphoma, 14 cases (5.6%) of sarcoidosis and 1 case (0.4%) each of sarcoma and mesothelioma were seen.
Conclusions EBUS guided TBNA, in our experience, is a reliable technique for staging of lung cancers, and investigation of unexplained mediastinal lymphadenopathy and masses. LBC reduces the number of slides from 7.5 per case to 1 slide, reducing time required for diagnosis.
- endoscopic ultrasound
- lung cancer
- lymph node pathology
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
In February 2008, the National Institute of Clinical Health and Excellence (NICE) issued an interventional procedure guidance document, recommending the use of endobronchial ultrasound (EBUS) guided transbronchial fine needle aspiration (TBNA) for mediastinal masses.1 EBUS guided TBNA is increasingly used worldwide, and in the UK, for the accurate staging and classification of lung cancers.2 As per NICE guidance on lung cancers, where appropriate, a histopathological or cytological confirmation of mediastinal lymph node involvement by lung cancer should be performed routinely, to determine the treatment protocol of patients with lung cancers.3 EBUS is increasingly preferred over mediastinoscopy or thoracoscopy for the sampling of mediastinal lymph nodes.4 Both these interventional procedures require general anaesthesia, and can be associated with complications. EBUS has the advantage of being safe, minimally invasive, and can be done as an outpatient procedure, using a local anaesthetic under conscious sedation. Only the upper anterior mediastinal lymph nodes are accessible by mediastinoscopy. EBUS can access a range of anterior mediastinal lymph nodes. The main indication for EBUS is lung cancers suspected to have metastatic disease on CT/PET scan. EBUS guided TBNA is also used in the investigation of unexplained mediastinal lymphadenopathy and masses due to other causes, more commonly granulomatous inflammation-like sarcoidosis.4
In our institution, this technique was implemented in January 2008. Initially, a cytology team were present at the time of the procedure, for assessment of adequacy of material obtained for diagnosis. Conventional smears were prepared. Subsequently, with more experience of the procedure, it was less necessary for cytology staff to be present to assess adequacy. This was also combined with the use of the Cytyc T2000 processing unit for cytology. The sample was sent in the provided phial with CytoLyt preservative solution, eliminating slide preparation by the clinical staff. Normal saline needle washings constituted a main part of the diagnostic material. Cell blocks were prepared for histology and immunohistochemistry. This paper reports our experience with the use of EBUS guided TBNA at the University Hospital of North Tees hospital over a period of 20 months. Criteria for adequacy, the use of ThinPrep liquid based cytology and the pathology of mediastinal lymphadenopathy and masses are discussed.
In our institution, the usual clinical indication for EBUS guided TBNA of mediastinal lymphadenopathy is for the staging of lung cancers. As part of a centralised service for EBUS guided TBNA, our institute also received cases from other hospitals in the County Durham and Tees Valley region. Those nodes that were significantly enlarged on CT scan, with or without PET scanning, were sampled. PET positive nodes, regardless of the size on CT scan, were also sampled. Unexplained lymphadenopathy or extraluminal tumours inaccessible by the conventional methods, such as bronchoscopy, were also investigated by EBUS guided TBNA. The nodes that are easily accessed by EBUS, are upper paratracheal (R2, L2), prevascular and retrotracheal (station 3), lower paratracheal including azygos (R4, L4), subcarinal (station 7), hilar (R10, L10) and interlobar (R11, L11). In our study, the nodes that were most frequently sampled included the subcarinal, lower paratracheal including the azygos nodes, hilar and interlobar mediastinal nodes. Up to three punctures, with up to 20 passes at each puncture were made for each node sampled. The tissue obtained, along with normal saline syringe washings, was submitted in a phial containing CytoLyt preservative solution. The sample was processed by the T2000 processor (Cytyc UK, Crawley, UK). One Papanicolaou stained slide was prepared. The remaining tissue was spun down into a pellet. Thrombin and normal human control plasma were used to induce clot formation. The clot was suspended in 1% agar and 10% formalin, and processed by histology. One H&E stained slide was prepared. Immunohistochemistry and any other special stains were requested as necessary. The cases were reported by the first five authors (SN, JH, MS, CH and JS).
Before May 2008, the samples were processed conventionally; air dried Giemsa stained smears and wet fixed Papanicolaou stained smears were made. A cytology team consisting of SN and one BMS staff member were present to assess adequacy. Initial air dried smears were stained by the DiffQuik method, and assessed for cellularity. Further samples were requested if there was no evidence of lymph node sampling, or only necrotic material was obtained. Over a period of 2 months (approximately 15 procedures), the presence of the cytology team was required less at the time of the procedure. In May 2008, Cytyc T2000 was procured by the laboratory; this eliminated the process of slide preparation by the clinical staff.
A total of 250 specimens from 229 patients were included in the study (table 1). There were 132 male patients and 97 female patients, aged 19–87 years. The indication for the procedure was mainly staging of lung cancer. In a smaller number (19 cases), the procedure was also used for diagnosis of unexplained mediastinal lymphadenopathy or mass. When the conventional method was used to process the material, an average 7.5 slides per case were received (range 1–14 slides).
Needle washings were processed separately for cell block in 85.2% of the total cases (table 2). However, in 17 cases (35.4%) processed conventionally, 14 cases (48.3%) processed by both LBC and conventionally, and 30 cases (17.1%) processed by LBC, adequate diagnostic material was not present on the H&E slide.
Immunostain requests on cell block material were made in the following cases: small cell carcinoma (n=18); non-small cell carcinoma (n=23); others, thyroglobulin and TTF1 (n=1); and lymphoma panel (n=3). In seven cases, ZN stain was requested to exclude tuberculosis.
EBUS guided TBNA cytology is increasingly becoming the procedure of choice to stage lung carcinoma.5 The procedure itself is time consuming, but in experienced hands is accurate. The diagnostic yield, as in any other field of cytology, is mainly dependent on the aspirator or the person performing the procedure.6 The diagnostic accuracy can reach up to 97%, according to some reports.7 8
There appears to be a learning curve with regard to the procedure. The inadequate rates are higher at the beginning of the learning curve. With more experience with the procedure, the inadequate rates have fallen from 16.6% to 9.8%. In our experience, adequacy is independent of the use of Cytyc T2000, as the rates are similar when the material was processed conventionally, and during the phase when both conventional cytology slides and LBC were used for processing (16.6% and 17.2%, respectively). At the time of implementation of EBUS, a cytology team was present to assess adequacy. This was felt to be essential to help the clinician gain experience. It also helped the reporting cytopathologist understand the process, and the cellular content of the aspirate, helping interpretation. There was a steep learning curve with regard to the reporting of specimens. Multiheader reviews of cases helped in the diagnostic interpretation. The learning curve for EBUS TBNA for thoracic clinicians is reported to be approximately 10 procedures.9 In our institution also, the presence of cytology staff was less of a requirement after a period of about 2 months (approximately 15 procedures), as more adequate samples, including microbiopsies, were obtained. In a busy cellular pathology department, it may not be possible to provide a cytology team for assessment of adequacy, because it can take up to 45 min for a single procedure. In May 2008, our department procured the Cytyc T2000, making the processing of the material easier.
With the conventional method, the pathologist would have to examine between 1 and 14 glass slides, which is laborious. The time taken would average between 45 min and 90 min. With the implementation of the T2000, the pathologist examined 1 ThinPrep slide, along with cell block material. The time taken was reduced considerably, between 20 to 45 mins per case. Introducing the T2000 also helped in the preservation and fixation of the material. The problems of air drying and fixation artefacts were eliminated by liquid based cytology. In our experience, these were the biggest advantages of the T2000 processor.
Needle syringe washings are an essential part of the procedure, as they improve diagnostic yield. Needle washings were submitted separately in the initial aspirates. Very often, the needle washings were found to be the more cellular part of the aspirate. We have asked the clinicians to submit both parts of the aspirate together in one phial.
Needle washings were processed separately as cell block material in 85.2% of the total cases. However, in 17 specimens (35.4%) processed by the conventional method, no diagnostic material was present on the H&E slide. Of these, eight cases were inadequate, eight cases were adequate but showed no malignant cells, and in only one case was a diagnosis of small cell carcinoma made on the prepared slides. Similarly, in 14 specimens (48.3%) processed by both LBC and conventionally, no diagnostic material was present in the cell block. Of these, five cases were inadequate, eight cases showed no malignant cells but were adequate, and in one case a diagnosis of atypia suspicious for a non-small cell carcinoma was made. In 30 specimens (17.1%) processed by T2000, either there was no material sent for cell block or no diagnostic material was present on the H&E slide. Of these, 17 cases were inadequate, nine cases showed no malignant cells but were adequate, in three cases a malignant diagnosis was made, and in one case a diagnosis of atypia suspicious for non small cell carcinoma was made. Overall, in 35–56% of those cases with no cell block, the aspirates were inadequate. It is possible that with increasing experience with the technique, we received more adequate diagnostic material.
For assessing adequacy, the presence of lymphoid material was essential (figure 1b). A sample can be considered to be adequate only when it contains abundant lymphocytes, either dispersed, or in the form of lymphoid aggregates and follicles (figure 1d). The presence of anthracotic pigment within macrophages was corroborative evidence of having hit the lymph node. Free lying anthracotic pigment was less significant, and could be of bronchial mucosal origin. Bronchial epithelial cells were often present in large numbers (figure 1a). Their presence was considered evidence of traversing the bronchial mucosa, but was not accepted as evidence to call a specimen adequate. In several aspirates, bronchial cartilage was also present. One has to be aware of the various structures that are traversed during the course of aspiration to avoid misinterpretation.10 In 28.8% of the total cases, the material consisted of reactive lymphoid tissue with no malignant cells seen. Reaspiration was performed in eight cases. In only one case did reaspiration show metastatic large cell carcinoma. The overall inadequate rate was 12%. In one of the inadequate cases, only necrotic material was present, and the sample was deemed inadequate. Reaspiration was performed in seven inadequate cases; three were adequate with no metastases on reaspiration. One case showed amyloidosis.
It is not within the scope of this paper to provide a true comparison between the two processing techniques with regard to diagnostic adequacy and accuracy. There is no standardisation of data with regard to patient characteristics, number of passes, nodal station sampled, node size and experience of the person performing the technique. The use of Cytyc T2000 did, however, eliminate the need to prepare slides by the clinicians, reduce the number of slides required for diagnosis, and eliminate the diagnostic problems associated with fixation artefacts.
In 48.8% of the cases, the lymph nodes showed evidence of metastatic lung carcinoma. The malignant cells were obtained from small cell and non-small cell carcinoma (figures 2c,d, 3a–d). CK7, CK20 and TTF1 immunostains were used to confirm primary site of origin. PanCK, Cam5.2, CD56, Synaptophysin, Chromogranin, CK34 (34bE12) and LCA were used as a panel for small cell carcinoma (figure 2a,b). In only one case the material was crushed; malignant cells present were difficult to interpret. Immunohistochemistry was equivocal, and not helpful in distinguishing between a small cell and non-small cell carcinoma. The bronchial biopsy showed a small cell carcinoma. In five cases, a diagnosis of atypia, suspicious for non-small cell carcinoma, was made. In one case of known squamous carcinoma of lung, scanty atypical squamous cells were present in a background of lymphoid cells. Reaspiration confirmed the presence of metastatic disease. One needs to be aware that in the absence of evidence of lymph node sampling, atypical squamous cells present could be from bronchial mucosa showing squamous metaplasia. In such cases, a diagnosis of metastatic squamous cell carcinoma needs to be made with caution. In one case with known oral carcinoma, review of cytology showed markedly reactive bronchial epithelial cells, and the case was regarded as reactive and not suspicious for malignancy. Markedly reactive bronchial epithelial cells can be a diagnostic pitfall, however the presence of cilia, other unremarkable bronchial epithelial cells in the vicinity, usually help in avoiding misinterpretation. In this case, reactive bronchial epithelial cells were present in the LBC preparation and only blood was seen in the conventional smear. There was no further reaspiration. An independent review of the remaining three cases of atypia suspicious for malignancy revealed malignant cells from non-small cell carcinoma. In one of these cases, processed both conventionally and by LBC, showed sufficient material on both preparations to make a diagnosis. There was no reaspiration. The other two cases were processed by T2000 only. One was confirmed to be a non-small cell carcinoma on mediastinoscopy and the other was treated as non-small cell carcinoma. Like the clinicians, there is a definite learning curve with the reporting of the specimens. A second opinion or multiheader review of such cases causing a diagnostic dilemma is recommended and is our usual practice.
Lymphoma (figure 4a–d) is an important differential diagnosis to be considered, in the diagnosis of unexplained mediastinal lymphadenopathy. Three cases of non-Hodgkin lymphoma were diagnosed, and the cell block was found to have sufficient tissue to enable the regional lymphoma diagnostic team to perform a complete lymphoma immunohistochemical panel, and categorise the lymphoma. One case was of low grade B cell lymphoma, most probably of mantle cell type. Bone marrow biopsy confirmed the diagnosis. One case was of recurrent high grade B cell lymphoma. The third case was suspicious but not diagnostic of B cell lymphoma, and further investigations were requested. A sensitivity of 90.9%, specificity of 100%, positive predictive value of 100% and negative predictive value of 92.9% have been reported in the use of EBUS guided TBNA in the diagnosis of lymphoma.11 One case of metastatic mesothelioma (figure 5a–d) and one case of poorly differentiated carcinoma/sarcoma were seen. The cell block in both cases contributed to the diagnosis, and immunohistochemistry was performed.
In 14 cases, mediastinal lymphadenopathy was due to sarcoidosis. Zeihl Neilson stain was negative. Granulomas, though better appreciated in the cell block, were also seen in the ThinPrep slide as ill defined aggregates of macrophages. In five cases, granulomas were seen only in the cell blocks (figure 1c). In one case, the ThinPrep slide showed granulomas; no cell block was prepared. In the remaining eight cases, both cell block and ThinPrep slide showed granulomas.
In those cases with unexplained mediastinal lymphadenopthy, part of the sample could be sent for microbiological culture studies to exclude tuberculosis in granulomatous inflammation. In cases of suspected lymphoma, samples could be sent for immunophenotyping with flow cytometry. These investigations were not done in our case series, but we hope to do these in the future.
In one case, the mass in the mediastinum was a multinodular, retrosternal goitre, which has been published as a case report.12 In our series, we had one case of mediastinal amyloidosis presenting with a mediastinal mass. The first aspirate was inadequate. Repeat EBUS guided TBNA showed amyloidosis. The amyloid material was seen in the cell block and not on the ThinPrep slide. Serum analysis showed monoclonal IgM Lambda paraproteinaemia. A re-biopsy of the mediastinal mass with oesophageal endoscopic ultrasound showed the same features as seen in the EBUS specimen. The amyloid present was classified as λ light chains. The case was reported to the National Amyloid Centre (figure 6a–d).
In summary, EBUS guided TBNA is an accurate, cost effective, minimally invasive procedure, that can be performed as an outpatient basis, and is increasingly becoming the mainstay in the staging of lung cancers. It can also be used for the investigation of mediastinal lymphadenopathy masses. Liquid based cytology is the preferred method for the processing of diagnostic material as it reduces the number of slides required for diagnosis, eliminates the need for slide preparation by clinicians and eliminates fixation artefacts. It is however beyond the scope of this paper to provide a true comparison between the two processing techniques as the data are not standardised. Saline washings improve diagnostic yield. Cell block for histology and immunohistochemistry provides supportive evidence for the diagnosis.
Endobronchial ultrasound guided transbronchial fine needle aspiration is the procedure of choice for staging of lung cancers and for the investigation of unexplained mediastinal lymphadenopathy and masses.
Liquid based cytology is the preferred method for the processing of diagnostic material as it reduces the number of slides, eliminates the need for slide preparation by clinicians and eliminates fixation artefacts. It is however beyond the scope of this paper to provide a true comparison between the two techniques as the data are not standardised.
Saline needle washings improve diagnostic yield. Cell block for histology and immunohistochemistry provides supportive evidence for diagnosis.
Cytology staff may be needed to assess adequacy in the initial stages of implementation of the procedure.
We would like to thank Mr J B Smith (Head BMS cytology), Ms P Clark (BMS cytology) and Ms S Coyle (BMS cytology) for their technical help and support.
Competing interests None.
Provenance and peer review Not commissioned; not externally peer reviewed.