Endoluminal or endoscopic ultrasound (EUS) was first attempted in 1957 by Wild and Reid who placed a mechanical ultrasound transducer in the rectum.1 It was not until 1975 that the upper gastrointestinal tract was examined when Rasmussenet al 2 measured the stomach wall thickness with a 6 MHz transducer passed through the biopsy channel of a gastroscope. In the 1980s, with the development of a dedicated endoscope incorporating a mechanical ultrasound transducer, EUS became important in clinical practice. Accurate local and nodal staging of oesophageal, gastric, and pancreatic tumours3-6 and assessment of stone disease in the biliary tract7 established EUS in the investigation of gastrointestinal disease. The accurate detection of mediastinal lymph nodes in oesophageal cancer had obvious implications for patients with lung cancer, and the role of EUS in lung cancer was first described in Japan in 1988.8 Further work confirmed the superior accuracy of EUS in the nodal staging of lung cancer compared with computed tomographic (CT) scanning,9 although EUS is not yet used routinely in the preoperative staging of lung cancer in the UK.

Further technical advancement led to the development of the linear EUS probe. This allows passage of a needle down the biopsy channel of the endoscope, through the wall of the gastrointestinal tract, and into adjacent structures such as lymph nodes. The orientation of the ultrasound beam, parallel rather than perpendicular to the long axis of the endoscope, allows continuous ultrasound monitoring of the needle tip. Several studies have shown that transoesophageal EUS guided fine needle aspiration (EUS-FNA) is a simple, relatively non-invasive method of obtaining tissue from various nodal stations in the mediastinum.10-12 Only the anterior mediastinum is off limits because of air in the trachea. It is performed as a day case in most patients, it is safe, and requires only conscious sedation. Complications related to EUS-FNA reported by Wiersemaet al in 457 patients with 554 lesions occurred in only three patients who underwent aspiration of cystic pancreatic lesions (haemorrhage and fever). This series included 192 peri-intestinal lymph node biopsies with no complications.13

Does the technique have a role? Sensitivities of 85–92%13 ,14 for malignancy in lymph nodes suggest that EUS-FNA is worth considering before more invasive techniques. For example, EUS-FNA can be used before mediastinoscopy where bronchoscopic biopsy specimens have failed to provide a tissue diagnosis. Fritscher-Ravens et al describe 16 patients with an intrapulmonary lesion on the chest radiograph or CT scan who had inconclusive pathology results obtained at bronchoscopy with forceps biopsy and/or brush cytology.15 Of the 10 patients with a final diagnosis of malignancy, this was established with EUS-FNA in nine. Transcarinal needle aspiration was not performed in this study which might have obviated the need for EUS-FNA in some of these patients.

A more systematic approach to the role of EUS-FNA has been explored as part of a routine staging protocol for patients with known or presumed lung cancer.10 ,12 Gress et al 12 reported that EUS-FNA avoided unnecessary surgery in 14 of 24 patients, confirming N2 disease in two and N3 disease in 12 patients. Mediastinoscopy, altogether more invasive and expensive than EUS-FNA, was avoided in all 24 patients with only one false negative result. Aabakken et al have also shown that EUS-FNA is a cost effective alternative to mediastinoscopy/mediastinotomy in a comparison using a cost effectiveness model.16 However, there are few proper comparative data between the techniques. Despite small patient numbers, in the only comparative study of EUS-FNA with mediastinoscopy to date Serna et al reasonably concluded that the techniques may be complementary, mediastinoscopy targeting the upper and anterior mediastinum and EUS-FNA targeting subcarinal and posterior mediastinal lesions.17 Further comparative data are required to assess properly its role as part of formal lung cancer staging protocols but the initial pointers are favourable.

Let us briefly consider some of the other relatively non-invasive techniques for obtaining tissue. Percutaneous needle biopsy with CT guidance is an alternative which was reported to have a remarkably high sensitivity of 98% (40 of 41) for diagnosing carcinoma in one series,18 although this was in a selected population thought to have lung cancer. In the same series, however, pneumothorax occurred in 34% of patients with a chest tube being required in 14%. Another limiting factor in this study was that biopsy samples were only taken from nodes with a diameter greater than 1.5 cm, although biopsy samples were successfully taken from nodes with a diameter of 1.2 cm in another series19 which again had a significant pneumothorax rate of 22%. In our experience nodes of less than 1 cm in diameter can be aspirated relatively easily and without complication with EUS-FNA.20 Although there are disadvantages to the percutaneous approach, particularly the higher complication rate, a major advantage is that no specialist equipment is necessary as virtually all district hospitals have a CT scanner. This, of course, assumes that the radiological skill, inclination, and time to take biopsy samples is available.

Transbronchial/carinal needle aspiration (TCNA) performed at the time of flexible bronchoscopy has a variable yield and sensitivities ranging from 34% to 89%.21-23 More recently there have been preliminary reports of the procedure being assisted by CT scanning or endobronchial ultrasound. With CT guidance the sensitivity increased from 20% to 60%24 and, on a per node basis, a sensitivity of 88% was achieved when guided by virtual bronchoscopy.25 Shannon et alreported that EUS guidance reduces the number of passes required, but does not increase the already high yield without guidance.26 The sensitivities with and without EUS guidance were 82% and 90%, respectively, in this series with on site cytopathology. One of the problems with TCNA is that the promising results obtained in academic centres have not necessarily been repeated in smaller units. This variability in yield may have prevented the more widespread use of the technique. When used for staging lung cancer, concerns about TCNA have also been raised regarding false positive results caused by contamination of the specimen with malignant cells from the lumen.27 ,28 This should not be a problem with the transoesophageal approach.

In conclusion, the role of EUS-FNA in assessing mediastinal pathology needs greater attention. It is clearly a useful technique and the tissue obtained safely and relatively non-invasively will influence patient management in the majority of cases. There are no more than five centres performing this procedure in the UK; three years ago there was only one. For a proper assessment of its role in preoperative staging of lung cancer, more centres need to perform the technique which is relatively simple to learn and appropriate comparative studies could then be set up. EUS-FNA was primarily developed with gastrointestinal disease in mind, particularly staging and obtaining tissue from pancreatic cancer. Biopsy specimens can also be taken from lymph nodes adjacent to the gastrointestinal tract below the diaphragm. The indications have expanded further with EUS guided pancreatic pseudocyst drainage and coeliac plexus neurolysis for pain relief. The ultrasound machine required (Hitachi) can be used for general ultrasound work and the cost of setting up this service may be spread by consultation with gastrointestinal and radiological colleagues. EUS has moved on considerably from the rectum since 1957 and there is now adequate evidence for an increased role in the management of a number of diseases across several specialities.