Endoscopic ultrasound: an overview of its role in current clinical practice

Introduction

Endoscopic ultrasound (EUS) combines endoscopic visualisation of the gastrointestinal tract with high frequency ultrasound within the structure of a modified endoscope. The merging of these two technologies enables precise imaging of the layers of the gastrointestinal wall as well as accurate assessment of extraluminal structures, thereby facilitating therapeutic intervention. This review outlines the diagnostic and therapeutic applications of EUS, with comparisons to conventional techniques.

Equipment

Echoendoscopes are designed using either a radial or curvilinear array system. The format may be mechanical or electronic. The electronic echoendoscope is now more favoured as it contains no moving parts and is thus more durable. The design is essentially that of a modified gastroscope, having both optical video views as well as ultrasound capability.

Radial

Mechanical radial echoendoscopes were available commercially in the late 1980s. A rotating ultrasound transducer with a range of frequencies between 5–20 MHz ¹ is situated distal to an oblique‐viewing lens at the tip of the endoscope. A water filled balloon allows for acoustic coupling (Fig. 1). The images obtained are cross‐sectional in nature, perpendicular to the endoscope shaft, akin to ‘slices’ obtained via CT scanning (Fig. 2a). Electronic radial echoendoscopes provide Doppler capabilities.

Fig. 1.

The electronic radial echoendoscope (Olympus Corporation, Tokyo, Japan).

Fig. 2a.

The radial echoendoscope scans at an axis perpendicular to the endoscope shaft.

Linear

The scanning plane of electronic linear echoendoscopes is oriented in the same plane as the scope shaft and accessory channel with the field of view ranging between 120° to 180°. The most important difference between radial and linear echoendoscopes is the ability to perform fine needle aspiration using the linear echoendoscope. Fine needle aspiration (FNA) cannot be performed using the radial echoendoscope because the ultrasound beam would pass through the needle at right angles and the needle would appear as a ‘dot’. With the linear echoendoscope, however, the needle passes in the same axis as the ultrasound beam, thus it is visible in its entirety as it is passed into the targeted lesion (Fig. 2b).

Fig. 2b.

The linear echoendoscope scans in a plane parallel to the endoscope shaft. Instruments inserted through the accessory channel are visualised as they pass through the ultrasound beam.

FNA needles

FNA needles for EUS applications range in size from 25G to 19G. Larger needles may increase trauma and result in a more bloody sample but are required for therapeutic EUS procedures where guidewires must be passed through the needle interior. Needles may have beveled or ball‐tips (the latter reduces the risk of scope channel injury during inadvertent deployment) and contain stylets, which prevent obstruction of the needle with “contamination” by normal gut wall mucosa as it is advanced through this layer into the lesion. Suction may be applied to aid aspiration of tissue. An EUS nylon cytologic brush is useful in sampling pancreatic lesions, where needle aspirates are often acellular ² .

19G Trucut biopsy needles are cutting needles that obtain core specimens, being potentially more accurate than EUS‐FNA for the evaluation of submucosal lesions and lymphomas ³ . These devices are technically demanding however, and do not function well when the echoendoscope is angulated, particularly in the second part of the duodenum ⁴ .

Technique

EUS procedures are performed in the same fashion as standard endoscopic examinations. The majority of cases are performed on an outpatient basis and intravenous sedation is usually employed. Procedure duration varies according to the complexity of the region being imaged and whether or not FNA is performed.

The echoendoscope is passed through the mouth until the tip reaches the target region. If the lesion of interest lies within the gut wall, water can be instilled into the gut lumen and the echoendoscope “floated” next to the lesion so that high quality images can be obtained using water as a conductive medium. Alternatively, acoustic coupling with the mucosa is achieved using a water‐filled balloon at the tip of the echoendoscope.

Extraluminal lesions are assessed using anatomical “stations”. The upper retroperitoneum (pancreatic body and tail, spleen, retroperitoneal lymph nodes, left adrenal gland and left lobe of the liver) is viewed through the gastric wall (Fig. 3). To assess the lower retroperitoneum (pancreatic head, common bile duct), the echoendoscope is positioned in the proximal duodenum (Fig. 4). Structures within the posterior mediastinum (heart, pleura, spine, vascular structures and posterior mediastinal lymph nodes) are visualised through the oesophageal wall (Fig. 5).

Fig. 3.

Visualisation of the pancreatic body and tail via the gastric wall (PD = pancreatic duct, BODY = pancreatic body parenchyma, SMV = superior mesenteric vein).

Fig. 4.

Views obtained through the duodenal cap: common bile duct closest to the transducer and pancreatic duct below this.

Fig. 5.

Mediastinal structures visualised via the oesophagus. (Ao = aorta, Az = azygous vein).

EUS‐FNA of mass lesions lying outside the gut wall is often performed by using the radial echoendoscope initially to identify the lesion and then using the linear echoendoscope to execute the actual biopsy. Colour Doppler enables the recognition of blood flow within vascular structures to ensure that no blood vessels lie between the needle and the targeted lesion. The needle is passed through the gut wall into the target lesion under real‐time ultrasound guidance (Fig. 6). The internal stylet is removed and the needle is passed back and forth with suction applied via a syringe. After withdrawal of the needle, the aspirated contents are expressed onto a slide or transport medium for cell block or flow cytometry. The presence of an on‐site cytopathologist to give instant feedback regarding specimen quality improves diagnostic certainty ⁵ .

Fig. 6.

Fine needle aspiration of a mediastinal lymph node.

Complications

As most echoendoscopes are oblique viewing and have a longer, more rigid tip than conventional endoscopes, passage of the instrument through the oropharynx should be made with due care. Despite this, the incidence of perforation does not appear to be more frequent than during standard endoscopy ⁶ . Most complications with EUS occur during therapeutic applications where the overall complication rate for EUS‐FNA is between 1–2% ⁷ and include infection (particularly for EUS‐FNA of pancreatic cystic lesions), haemorrhage, pancreatitis and duodenal perforation. Infectious complications following EUS‐FNA of solid lesions or lymph nodes are rare and prophylactic antibiotics are not recommended ⁸ .

EUS training

The Australian Conjoint Committee for the Recognition of Training in Gastrointestinal Endoscopy requires that candidates complete a minimum of 200 EUS examinations unassisted under supervision including 100 examinations for gastro‐oesophageal lesions/tumours and 100 examinations for pancreatico‐biliary investigations. A minimum of 50 FNA examinations (25 or more of which must be pancreatico‐biliary) must be performed unassisted under supervision ⁹ .

Indications

Diagnostic/staging

EUS has an established role in a wide variety of applications, including the assessment and staging of malignancy, evaluation of submucosal abnormalities, mediastinal lymphadenopathy and pancreatico‐biliary disease. EUS‐FNA compares favourably to US or CT guided percutaneous biopsy techniques particularly for smaller lesions ¹⁰ .

Malignancy

The American Joint Committee on Cancer stages luminal GI malignancies according to the TNM classification ¹¹ . A grade is given to depth of invasion (T), presence of locoregional lymph nodes (N) and presence of distant metastases (M). EUS is most beneficial in locoregional T and N staging, providing an accuracy of approximately 85% in GI luminal cancers ¹² .

In the assessment of cancers arising from within the gastrointestinal tract wall, EUS is of benefit as it can depict the five histologic layers of the gut wall in fine detail (Fig. 7), permitting accurate T staging. Extraluminal tumours that lie in close proximity to the gut lumen, such as pancreatic tumours, can also be staged with regards to invasion into nearby vasculature and other adjacent structures. EUS also allows visualisation of regional lymph nodes that lie adjacent to tumours and FNA can be performed where appropriate. During lymph node staging, the biopsy needle should not traverse the primary tumour as this may lead to false positive results. The impact of EUS‐FNA is significant in that it changes the management of patients with gastro‐intestinal, pancreatic and pulmonary malignancy, often resulting in the avoidance of unnecessary surgery ^{13 – 15} .

Fig. 7.

The five layers of the gastric wall at EUS, from inner to outer: innermost two layers (white and black) = mucosa, third layer (white) = submucosa, fourth layer (black) = muscularis propria, fifth layer (white) = serosa.

Oesophageal cancer

EUS has been demonstrated to have higher sensitivity in detecting oesophageal cancer when compared to CT and positron emission tomography (PET) with 18F‐Fluorodeoxyglucose ¹⁶ (Fig. 8) and represents an important complementary test in this setting. Using EUS with high frequency miniprobes, patients identified to have tumour limited to the lamina propria of the mucosal layer are unlikely to have lymph node involvement and may be candidates for endoscopic mucosal resection rather than oesophagectomy ¹⁷ . The accuracy of EUS‐FNA in detecting involvement of locoregional lymph nodes is >85% when compared to surgical specimens ¹⁸ . Most importantly, the presence of coeliac axis lymphadenopathy represents metastatic disease and EUS has been demonstrated to have superior sensitivity to CT and PET scanning in this setting ¹⁶ .

Fig. 8.

Staging of oesophageal cancer: the hypoechoic expansion of the oesophageal wall with tumour (T) does not invade the muscularis propria (black layer), hence is staged at T2. There is a regional lymph node (LN) measuring 0.73 × 0.5 cm that is round and hypoechoic, suggestive of malignant involvement. The aorta (AO) and heart can be seen adjacent.

In patients with Barrett's oesophagus and high‐grade dysplasia, the role of EUS for the detection of occult cancer and malignant lymphadenopathy is yet to be clearly defined ¹⁹ .

Gastric cancer

EUS is beneficial in the assessment of early gastric cancer. Those patients without submucosal invasion (T1) can be considered for endoscopic mucosal resection rather than gastrectomy. For established gastric cancer, EUS is superior to CT in assessing locoregional stage ²⁰ , although three dimensional multidetector row CT techniques have improved accuracy ²¹ . In patients with gastric lymphoma, EUS is particularly accurate in assessing T stage ²² . EUS‐FNA is valuable for lymph node sampling and biopsying the gastric wall when the EUS appearance is abnormal but mucosal biopsies are non‐diagnostic.

Rectal cancer

In patients with rectal cancer, neoadjuvant and adjuvant chemoradiation therapy is indicated for advanced locoregional disease. EUS is between 80–95% accurate and is superior to CT for T staging ²³ . Staging using MRI with rectal coils has similar efficacy to EUS, except in the differentiation between T1 (invading submucosa) and T2 tumours (invading muscularis propria), where EUS may be superior ²⁴ . Accuracy of EUS is reduced following radiotherapy, due to the presence of inflammation and fibrosis ²⁵ . EUS accuracy in N staging is similar to CT and MRI as benign inflammatory lymphadenopathy may accompany rectal cancer ²⁶ .

Pancreatic cancer

The sensitivity of EUS for the detection of a pancreatic mass was 96% when results from 22 studies were combined ²⁷ . However, when benign lesions and ampullary tumours were excluded, sensitivity decreased. Comparisons with helical CT, angiography, MRI and PET suggest that EUS is more sensitive for the detection of tumours and vascular invasion (Fig. 9) but that CT, MRI and PET are complementary for the determination of resectability of the tumour ^{28 – 31} . The sensitivity and specificity of EUS‐FNA for the diagnosis of pancreatic tumours is 85% and 98% respectively ²⁷ . EUS‐FNA of pancreatic cancer has gained favour due to the risk of needle track seeding with percutaneous biopsies ³² . The tissue planes that are passed when performing transduodenal EUS‐FNA for a pancreatic head cancer will be resected in any subsequent surgery, thus the risk of seeding via this technique is inconsequential.

Fig. 9.

A 1.75 × 1.78 cm tumour (‘mass’) in the head of pancreas impinges upon, but does not invade, the portal vein (‘pv’).

In the localisation of pancreatic neuroendocrine tumours, which are often <1 cm, EUS is superior to CT, MRI and somatostatin receptor scintigraphy ³³ (Fig. 10). EUS is also superior to CT, MRI and transabdominal US for the staging of periampullary carcinomas ³⁴ .

Fig. 10.

A neuroendocrine tumour in the body of the pancreas, FNA being performed. Note the sharp bordered, small, rounded configuration with a combination of cystic and solid elements – this is typical of the EUS appearance of neuroendocrine tumours.

Pancreatic cystic tumours may be benign, malignant or have malignant potential and differentiation between these types using conventional imaging is difficult. EUS can be considered complementary for distinguishing such lesions ^{35 , 36} (Fig. 11) although one study found little interobserver agreement ³⁷ . EUS‐FNA sampling of cystic fluid (Fig. 12) distinguishes mucinous from non‐mucinous cysts by measurement of cyst fluid CEA levels with high specificity but does not predict malignancy ³⁸ .

Fig. 11.

A unilocular cystic lesion (‘CYST’) measuring 2.28 × 2.58 cm. Note that it communicates with the main pancreatic duct (‘PD’), suggesting a diagnosis of intraductal papillary mucinous tumour.

Fig. 12.

The clear, non‐viscous fluid aspirated at EUS from a serous cystadenoma.

Submucosal GIT lesions

Submucosal lesions are often encountered during routine endoscopy, mucosal biopsies of which are often non‐diagnostic. EUS determines the layer of origin and detects characteristic appearances of cysts, lipomas, leiomyomas and gastrointestinal stromal tumours (Figs. 13,14). EUS‐FNA of such lesions has a diagnostic yield of up to 91% ⁴⁴ .

Fig. 13.

A gastrointestinal stromal tumour (GIST). It has a characteristic hypoechoic appearance (T) and arises from the fourth layer of the gastric wall, the muscularis propria (MP).

Fig. 14.

A lipoma arising from within the gastric wall. Endoscopically, this would look the same as the GIST seen in Fig. 13. However, EUS allows differentiation between the two: the lipoma is brightly hyperechoic.

Lung cancer

EUS‐FNA is >90% accurate in nodal staging of non small cell lung cancer (NSCLC) ³⁹ , being more sensitive than CT ^{40 , 41} and more specific than PET ^{40 – 42} . This investigation also has efficacy in assessing tumour stage, biopsy of tumour adjacent to the oesophagus and assessment of metastatic disease in the left liver lobe and left adrenal gland. EUS‐FNA reduces the need for mediastinoscopy/thoracotomy by up to 50% ⁴³ and should be considered the first line investigation for tissue sampling of nodes in the posterior mediastinum (Fig. 15).

Fig. 15.

A large malignant lymph node seen at the 11 o'clock position within the mediastinum of a patient with a known primary lung carcinoma.

Mediastinal lymphadenopathy of uncertain aetiology

Posterior mediastinal masses (aortopulmonary window, subcarinal and perioesophageal stations) are usually initially detected on CT and have a wide differential, including infective and granulomatous disease, lymphoma, primary pulmonary and metastatic malignancy. These masses in the posterior mediastinum are readily amenable to EUS‐FNA via a transoesophageal route. EUS visualisation of anterior mediastinal lymphadenopathy is disrupted by air interference from the trachea; these groups are better assessed with endobronchial ultrasound.

When all four features: abnormal size >1cm, hypoechoic appearance, round shape and distinct margin are present, there is a high specificity, but low sensitivity for malignant infiltration ^{45 , 46} . EUS‐FNA provides a specimen sufficient for interpretation in over 95% of cases ⁴⁷ with a sensitivity of 96% for the detection of nodal malignancy in patients with known malignant disease ⁴⁸ . The American Society for Gastrointestinal Endoscopy recommends that FNA be performed during EUS evaluation of mediastinal lymphadenopathy, when the result will alter management ⁴⁹ .

Benign pancreatic disease

The diagnosis of chronic pancreatitis is often difficult to establish on conventional imaging with CT, abdominal US and ERCP. EUS can be used to detect characteristic alterations of the pancreatic parenchyma and duct although there exists strong operator dependence. Hence, the role of EUS is complementary to other modalities in this setting. Autoimmune pancreatitis has characteristic EUS appearances and use of EUS‐FNA increases the diagnostic yield ⁵⁰ .

Biliary stones

EUS has the advantage of visualising the biliary tree from within the duodenum without interference from abdominal gas or fat (Fig. 16). When compared to ERCP or intraoperative cholangiogram findings in a population of patients clinically suspected to have choledocholithiasis, EUS has a sensitivity between 89–94% and a specificity of 94% ^{51 , 52} . A meta‐analysis of five randomised, prospective, blinded trials comparing EUS and MRCP found them to be of comparable sensitivity and specificity ⁵³ . A cost‐benefit analysis found EUS to be of greatest value in the setting of intermediate (11%–55%) risk for choledocholithiasis. However, ERCP remains preferable for patients whose pre‐test probability is high (>55% risk) because therapeutic intervention can be performed simultaneously ⁵⁴ .

Fig. 16.

The CBD can be viewed in its entirety from the duodenal cap.

Perianal disease

Rectal EUS has been found to be effective in the assessment of perianal diseases. In the assessment of perianal Crohn's disease, EUS has a 91% accuracy, comparable and complementary to MRI and evaluation under general anaesthetic ⁵⁵ . EUS also has >90% sensitivity for the detection of anal sphincter defects in faecal incontinence ^{56 , 57} however studies comparing EUS with MRI are conflicting ^{58 , 59} .

Therapeutic EUS

The echoendoscope can be sited adjacent to extraluminal structures, facilitating therapeutic injection and drainage procedures with great accuracy.

EUS‐guided injection therapy

Coeliac plexus block is achieved by the injection of bupivacaine and neurolysis with the injection of absolute alcohol. A linear echoendoscope is directed towards the coeliac ganglia at the origin of the coeliac trunk with a 22–19G FNA needle. Durable analgesia is obtained in up to 91% of patients with pancreatic cancer ^{60 , 61} . Complications are rare (1% ⁶² ) and the EUS technique is safer than the CT guided percutaneous approach ⁶³ . The procedure is less efficacious in chronic pancreatitis however; Gress, et al. demonstrated that neurolysis with corticosteroids (triamcinolone) reduced pain scores beyond 12 weeks in 26% of subjects ⁶⁴ .

The poor prognosis of pancreatic malignancy has prompted the use of EUS for intratumoural injection of chemotherapeutic agents ⁶⁵ . EUS guided fine needle injection (FNI) of adenoviral vectors targeting tumour cells has been described ^{66 , 67} . Animal studies have assessed the efficacy of paclitaxel injection via EUS‐FNI into the porcine pancreas ⁶⁸ . This modality of treatment remains in its experimental phases, however.

Attempts have been made at ablating the epithelial lining of cystic tumours of the pancreas with ethanol lavage via EUS‐FNI but with limited success ⁶⁹ . A preliminary study found that the combination of ethanol lavage and paclitaxel injection was safe and effective ⁷⁰ .

EUS‐guided drainage procedures

Surgical and percutaneous approaches to pancreatic pseudocyst drainage are associated with significant morbidity and mortality ⁷¹ . EUS‐guided transmural drainage of pancreatic pseudocysts is minimally invasive and does not result in problems such as cutaneous fistulae. It is performed by transgastric/transduodenal puncture under EUS guidance, followed by the insertion of double pigtail stents through the gut wall to create a cystogastrostomy/cystoenterostomy (Figs. 17,18). The procedure was successful in 94% of cases with no mortality in one series of 51 patients ⁷² .

Fig. 17.

A pancreatic pseudocyst. The FNA needle is seen passing from the gastric wall into the cyst under EUS guidance. Fluid can be aspirated and sent for culture. The tract between the stomach and cyst is then dilated and stents inserted.

Fig. 18.

The insertion of double pigtail stents over guidewires from the stomach into the pseudocyst to allow drainage of the cyst into the stomach.

Similar techniques have been reported for the drainage of mediastinal ⁷³ , hepatic ⁷⁴ , splenic ⁷⁵ , subphrenic ⁷⁶ and pelvic ⁷⁷ abscesses. EUS‐guided transmural cholecystenterostomy has been described in patients at high risk for surgical intervention ^{78 , 79} .

EUS‐guided cholangio‐pancreatic drainage following failed ERCP involves EUS‐guided puncture of a dilated pancreatic or biliary system, passage of a guidewire and insertion of a trans‐duodenal or trans‐gastric stent ⁸⁰ and can obviate the need for percutaneous transhepatic drainage in suitable patients.

Conclusion

EUS facilitates the diagnosis of GI luminal and extraluminal masses and allows staging of a variety of malignancies, resulting in the avoidance of unnecessary surgery. Such indications, in combination with an expanding number of therapeutic applications, have established the role of EUS as a safe, accurate and cost‐effective tool.