|Home | About | Journals | Submit | Contact Us | Français|
Since the joining of ultrasonographic and endoscopic technologies in 1980, there has been tremendous advancement in the use of endoscopic ultrasound (EUS). Initially EUS was used mainly to study the anatomic appearance of digestive organs. In the 1990s, with the development of the charge-coupled device, thinner, more flexible scopes were developed that simplified imaging for the endosonographer and improved patient safety. With the development of linear array EUS scopes, biopsies could be performed utilizing fine needle aspiration. Currently, several therapeutic techniques, which were formerly performed through radiologic or surgical procedures, are being safely done with EUS. These include celiac plexus neurolysis and endoscopic drainage of pancreatic cysts. With the advent of miniprobe technology, better delineation of the wall of the gastrointestinal tract can be obtained, facilitating endoscopic mucosal resection. Novel therapeutic uses for EUS are constantly emerging. This article describes the various extraintestinal applications of EUS that have been developed in the last 25 years. More gastroenterologists need training in order to utilize this challenging technology and improve patient access in the local community setting.
Endoscopic ultrasound (EUS) was developed in the 1980s. Its main use at that time was in the staging of gastrointestinal tumors. Initially used as a diagnostic tool, since the early 1990s it has been increasingly utilized in intervention and therapy. Its uses have broadened such that it now has indications for extraintestinal pathology. Use of EUS will continue to grow as new technology, such as the thin miniprobe, dual-plane reconstruction (DPR), intraductal probes, and the new electronic radial echoendoscope, become available. Over the next few years, EUS is foreseen as becoming one of the most powerful gastroenterological, radiological, and therapeutic instruments, catalyzing significant changes in patient management.
EUS has been shown to be safe1 and emerging studies indicate that it is cost effective when compared to modalities such as computed tomography (CT) scans, positron emission tomography (PET) scanning, magnetic resonance imaging (MRI), angiography, and surgery.2 Immunotherapy, radiofrequency ablation, injection of viruses, gene therapy, and DNA analysis could become established EUS applications in the near future and are currently at various stages of investigation.3
The radial scanning echoendoscope is the most commonly used endoscope. It has either a rotating mechanical transducer at the tip (Olympus Corp) or a fixed, phased-array electronic transducer (Pentax Corp). The mechanical transducer provides a 360-degree image, oriented perpendicularly to the shaft of the scope, whereas the fixed array produces a 270-degree view (Figure 1). The image provided is a cross-sectional view, similar to slices of a CT scan, and is therefore easier to interpret, especially for the novice. The ultrasound frequency ranges between 5 and 20 MHz. It should be noted, however, that higher frequency results in better resolution, but at the expense of depth of penetration, which is relationally decreased. Endosonographers usually use the 5 and 7.5 MHz frequencies to scan and the depth of penetration is usually around 5–6 cm. It should also be noted that Doppler cannot be performed with the mechanical transducer, but is possible with the electronic radial scope. Recently, an electronic, 360-degree radial scanning echoendoscope with Doppler capability and tissue harmonic imaging was introduced by Olympus. The endoscopic view of this scope is forward oblique. The key components of the transducer are the piezoelectric crystals that vibrate to produce ultrasonic waves. Acoustic coupling is usually achieved with a water-filled balloon around the instrument tip.
This echoendoscope provides a view parallel to the long axis of the scope, allowing for safe performance of EUS-guided fine needle aspiration (FNA; Figure 2). These scopes are manufactured by both Olympus and Pentax and employ electronic, fixed-array technology. Doppler is available, and very useful in differentiating vessels from tissue, especially for FNA. The scope can switch between 5 and 7.5 MHz frequency. The needles commonly used are 19 and 22 gauge. Linear scanning is generally more difficult to learn than radial scanning.
Miniaturization of the ultrasound transducer led to the development of the miniprobe. This high-frequency ultrasound probe (HFUP) or catheter is about 2–3 mm in diameter and is passed through the biopsy channel of a regular upper endoscope. The probe has a mechanical rotating transducer at the tip and frequency ranges from 12 to 30 MHz. Recently, DPR probes have been made available for surface and volume rendering and three-dimensional reconstruction. These probes provide detailed imaging of the wall of the gastrointestinal tract and precise depth of penetration of tumor. In contrast to the five-layered wall structure obtained with conventional EUS, HFUP can delineate up to nine layers. Dedicated miniprobes for intraductal ultrasonography (IDUS) are also available from 12.5 to 30 MHz.
Pancreatic adenocarcinoma is the fourth leading cause of cancer-related death in the United States. The 5-year survival rate after postoperative multimodality treatment is about 20%.4,5 In the 1980s and early 1990s, EUS was found to be the best modality to locally stage pancreatic cancer (Figures 3 and and44).6,7 With the advancement in technology of other radiological methods, especially multidetector triple-phase CT scanning, later studies report comparable sensitivities and specificities. Tumor staging accuracy ranges from 85% to 93% and nodal staging accuracy is approximately 72%.8
There have been four studies comparing EUS with helical CT for evaluating pancreatic cancer. Two of them9,10 found helical CT and EUS to be equivalent; the other two11,12 found EUS to be superior for tumor detection and staging. The discrepancies in the findings of these four studies have several causes and biases beyond the scope of this article. EUS and helical CT are complimentary. A helical CT should be employed for initial evaluation and EUS should be performed if no evidence of distant metastasis is found. EUS can also determine vascular invasion. EUS is less accurate in evaluating invasion into the superior mesenteric vessels, but is very accurate in diagnosing portal vein invasion.13 Nevertheless, with the advent of surgical portal vein resection, portal vein invasion does not necessarily preclude resection. Detection of vascular invasion by EUS is being challenged by advances in CT technology.
Multiplanar reconstruction of major vessels in proximity to tumors revealed a sensitivity of 50–75% and a specificity of 91–97% in a recent study.14 EUS is very sensitive in detecting small pancreatic tumors. Studies comparing EUS with helical CT suggest EUS is more sensitive for tumors smaller than 3 cm.15 Therefore EUS is also indicated when there is clinical suspicion of a pancreatic tumor and other imaging studies are either nondiagnostic or equivocal.
Studies comparing EUS and MRI suggest that EUS is more sensitive.16-18 Studies comparing angiography and EUS have varied results, but suggest that EUS is as accurate for determining vascular invasion, except invasion of the superior mesenteric vessels.18-20 A study comparing EUS and PET scanning found a sensitivity of 93% for EUS and 87% for fluorodeoxyglucose-PET.11 PET scanning could have a higher false-positive rate as inflammatory masses cause a positive scan. Problems could arise in differentiating an inflammatory mass from a malignancy in a patient with chronic pancreatitis. The specificity varies from 46% to 93%.21,22 Newer techniques such as contrast-enhanced Doppler imaging could result in improvement in sensitivity and specificity, as adenocarcinomas are hypoperfused and inflammatory lesions are hyperperfused.
The establishment of diagnosis can be accomplished by EUS-guided FNA. The complication rate is less than 1%.23 Multiple studies have demonstrated sensitivities between 75% and 90% and specificity between 94% and 100%, implying a false-negative rate between 10% and 25%. If EUS suggests resectability, EUS-guided FNA is not necessary, although this aspect remains controversial. Advantages of needle biopsy include alternative diagnoses such as lymphoma, islet cell tumors, and metastatic disease. The disadvantages include bleeding, pancreatitis, and tumor seeding, a first case of which was recently reported.24
Several studies have evaluated cost-effectiveness and concluded that EUS-guided FNA is a cost-effective strategy, with one study estimating a cost saving of $3,300 per patient based on avoidance of surgery.25,26
Neuroendocrine pancreatic tumors are rare tumors, with an incidence of less than 1 per 100,000 people. EUS is very sensitive in detecting these small tumors. Sensitivities vary from 82% to 93% and specificity is over 90%.27,28 EUS is comparable to somatostatin receptor scintigraphy (SRS) for detecting pancreatic gastrinomas, and more accurate for insulinomas due to the low density of SRS receptors. Both tests are clearly superior to CT, which has a detection rate of about 60% at best. Both techniques, however, miss a significant proportion of duodenal gastrinomas.29 Because there is a more frequent incidence of extrapancreatic tumors, EUS is frequently used in conjunction with other modalities such as SRS and triple-phase CT.
A recent study recommended screening for pancreatic cancer in familial pancreatic cancer kindreds. The diagnostic yield for detecting clinically significant pancreatic neoplasms was 5.3%.30 No recommendations are known for surveillance after initial screening.
There are five basic types of cystic lesions that are encountered in the pancreas.
Simple Cysts: These are usually small in size and are congenital. They are benign, with thin walls, filled with clear fluid, and without septation.
Serous Cystadenomas: These are usually benign, seen mostly in women in the seventh decade of life. These tumors tend to be located in the body and tail of the pancreas. Up to 10% of serous cystadenomas have central calcification, causing a characteristic sunburst pattern. They can also have multiple microcysts, forming a honeycomb pattern (Figure 5). FNA reveals low-viscosity fluid with low tumor markers and glycogen-containing cuboidal cells. If these tumors cause symptoms, such as abdominal pain, they can be resected; otherwise they can be left alone.
Mucinous Cystadenomas: These tumors have malignant potential, with approximately 25% harboring malignancy at the time of diagnosis. They are usually asymptomatic, but can cause abdominal pain, pancreatitis, diabetes, and jaundice. They are usually found in women in the fifth and sixth decades of life and are located in the body and tail of the pancreas. Mucinous cystadenomas contain multiple fluid-filled cavities with thin wall septae (Figure 6), the cavities being much larger (>1 cm) than the microcysts seen in serous cystadenomas. FNA is usually performed on the nearest and largest cavity, and the cavity is completely drained to prevent infection. Antibiotics are usually administered prior to FNA and for about a week after. Mucous and mucous-filled cuboidal or columnar cells are seen in about 30–50% of cases. The tumor marker that is particularly helpful in determining a mucinous tumor is the carcinoembryonic antigen (CEA) level in cyst fluid. A recent prospective study found that a CEA level of 192 ng/mL using the receiver-operating curve analysis provided a sensitivity of 75% and a specificity of 84%.31 All mucinous cystadenomas are potentially malignant and should be resected.
Intraductal Papillary Mucinous Neoplasm: These tumors arise from the epithelium of the pancreatic duct and have malignant potential. They are usually diagnosed in men in the sixth and seventh decades of life. Blockage of the pancreatic duct with mucous or growth accounts for symptoms of abdominal pain, pancreatitis, weight loss, diabetes, and steatorrhea. EUS reveals a dilated pancreatic duct, segmental duct dilation, or side branch dilation in the presence of normal echotexture of the pancreas. EUS-guided FNA of the dilated pancreatic duct can be performed. IDUS and pancreatoscopy are other methods that can be used to aid in diagnosis. The recommendation is limited resection, if possible, and avoidance of total pancreatectomy as recurrence rates are comparable after partial or total pancreatectomy for invasive disease. This avoids the serious metabolic consequences of a total pancreatectomy.
Pseudocysts: These benign cysts are inflammatory collections of fluid that stem from duct disruption. They are usually seen in patients with chronic alcoholic pancreatitis. Signs detected with EUS include a hyperechoic thick wall, lack of septations, and internal echoes representing debris (Figure 7). FNA usually reveals a fluid with low viscosity, high amylase levels, low tumor markers, and histiocytes. Pseudocysts can be drained using EUS as guidance, especially if they are symptomatic. The advantages of using EUS include gauging the distance between the cyst wall and the gut wall (which should be <1 cm) to ensure safe endoscopic drainage and searching for gut wall varices between the pseudocyst and the wall, which would preclude endoscopic drainage. With the advent of larger-caliber echoendoscope channels, it is now possible to place 8.5-French (Fr) stents into the cysts for drainage. A one-step drainage system with a preloaded 8.5-Fr straight stent has recently been developed for easier and faster EUS-guided drainage of pseudocysts.32 It is possible to dilate the opening with a 15-mm balloon 2–3 days later, enter the cavity with a forward-viewing scope, and remove necrotic debris using a Dormia basket.33 EUS-guided drainage of pseudocysts should be performed by experienced endosonographers in a tertiary care center with access to surgery and interventional radiology in the event bleeding or perforation occurs.
Normal pancreatic tissue is seen as a “salt and pepper”–type pattern on EUS. EUS is an excellent instrument in comparison to other imaging modalities for detecting changes of chronic pancreatitis because the transducer is in close apposition to the body of the pancreas from the greater curve of the stomach and the head of the pancreas from the duodenal bulb. It is very sensitive in detecting both parenchymal and ductal changes or criteria (Table 1 and Figure 8). Using receiver-operating curve analysis, a threshold of 3–4 criteria offers the best compromise between sensitivity and specificity, approximately 70%.34 EUS has been compared with endoscopic retrograde choliangiopancreatography (ERCP) and the secretin test. Catalano and colleagues35 showed that the secretin test had a 100% agreement with normal and severe pancreatitis by EUS criteria, but agreement was poor for mild and moderate disease. The agreement between ERCP and EUS criteria was excellent for normal, moderate, and severe disease, but poor for mild disease. Neither ERCP nor secretin tests are true gold standards.
There is also wide interobserver variability for the different EUS criteria. In a study reported by Wallace and associates,36 overall agreement for the diagnosis of chronic pancreatitis was 0.45 among 11 experienced echoendoscopists. The score for most individual criteria was less than 0.4 except for lobularity (0.51) and duct dilation (0.60). Pathology of chronic pancreatitis is not standardized and is usually patchy, but a surgical specimen would be the ideal gold standard for diagnosis of chronic pancreatitis. Hollerbach and coworkers37 combined EUS and FNA in patients with chronic pancreatitis and concluded that the addition of FNA increased the negative predictive value from 75% to 100% but only increased specificity from 60% to 67%. Therefore, improvements in sampling and analysis could help with the diagnosis of early chronic pancreatitis. Although EUS is extremely sensitive, it can overdiagnose chronic pancreatitis, and so it is best to combine a thorough history, especially alcohol usage, with at least 3–4 EUS criteria to make a firm diagnosis.
Autoimmune pancreatitis is an increasingly recognized benign condition and is very responsive to steroid therapy. The characteristic EUS findings for autoimmune pancreatitis include a diffusely enlarged, frankly hypoechogenic pancreas that lacks the features of regular chronic pancreatitis (lobularity, heterogeneous foci), or a solitary hypoechogenic mass in the head of the pancreas. EUS-guided FNA cytopathologic assessment of the aspirated material reveals evidence of chronic inflammation (lymphocytes and plasma cells) and fibrosis.38
EUS is the procedure of choice to evaluate patients with acute idiopathic pancreatitis. It is able to detect abnormalities such as microlithiasis, pancreatic and ampullary tumors, parenchymal changes of chronic pancreatitis, annular pancreas, choledochocele, and pancreas divisum (PD). In fact, duodenal aspiration of bile for crystal analysis after cholecystokinin administration can be performed with the echoendoscope if EUS findings are normal.39
PD occurs when the ventral and dorsal ducts of the embryonic pancreas fail to fuse during organogenesis. It is the most common congenital variant of pancreatic ductal development, occurring in approximately 10% of individuals. EUS has been found to be very accurate (94%) in the diagnosis of PD. Prominent dorsal pancreatic duct, cross-duct sign, and the absence of the stack sign are highly predictive of PD.40,41
Celiac plexus neurolysis can be safely performed using the anterior approach to the celiac plexus with a linear-array echoendoscope and a 19-gauge needle for injection. It is mainly performed to relieve pain of pancreatic cancer and chronic pancreatitis, although it is not very effective for the latter. About 10–20 mL of absolute alcohol with 25 mL of 0.25% bupivicaine is injected for pancreatic cancer and alcohol is replaced by 80 mg of triamcinolone for chronic pancreatitis. Alcohol is avoided in patients with chronic pancreatitis because of a small risk of serious complications and resulting fibrosis, making subsequent surgery difficult. Studies have shown a significant reduction in pain scores in close to 80% of patients, sustained at 24 weeks, for neurolysis in pancreatic cancer,42 but in only 10% of patients with chronic pancreatitis for the same time duration. A comparison of CT-guided versus EUS-guided neurolysis revealed that 50% of patients in the EUS group improved versus 25% in the CT group.43 Major complications are rare and minor complications include postural hypotension (20%), transient diarrhea (17%), and temporary worsening of pain (9%).44
EUS is very accurate in diagnosing choledocholithiasis and microlithiasis of the gall bladder (Figure 9). A prospective study comparing EUS, CT, and ultrasound revealed sensitivities of 97%, 75%, and 25%, respectively. The difference was more pronounced in the subgroup with nondilated ducts.45 Studies comparing EUS and ERCP have found them to be more or less equally accurate.46-48 Nevertheless, it is well-recognized that small stones may be missed on cholangiography, and ERCP is a more risky procedure. Canto and colleagues,49 in a comparative study of EUS and ERCP for diagnosing choledocholithiasis, stratified patients into low, indeterminate, moderate, and high risk for bile duct stones. There was no significant difference between the two procedures for overall accuracy. They suggested EUS as the most appropriate test, based on risk and cost, when the index of suspicion is low or uncertain.49
Studies comparing EUS and magnetic resonance cholangiopancreatography (MRCP) are limited. A prospective controlled study comparing the two revealed a sensitivity, specificity, positive predictive value, and negative predictive value of 100%, 95%, 91%, and 100%, respectively, for EUS and 100%, 73%, 63%, and 100%, respectively, for MRCP.50 Another study comparing the two, using ERCP as the gold standard, revealed EUS was more sensitive than MRCP for choledocholithiasis (80% vs 40%), although the number of patients was small.51 EUS is highly accurate in the diagnosis of microlithiasis with a 96% sensitivity and 86% specificity, and superior to bile analysis in one study.52 EUS is a very useful modality in the evaluation of idiopathic recurrent pancreatitis; a recent publication concluded that 50% of patients with intact gallbladder had microlithiasis.53 Choledocholithiasis can also be evaluated with IDUS. Comparison of IDUS with ERCP found IDUS to be superior (97% vs 81%).54
Patients with cholangiocarcinoma typically present with painless jaundice; standard imaging such as CT and ultrasound usually show biliary dilation but no identifiable cause. EUS has a sensitivity of about 85% for tumor-staging and about 90% for nodal staging if FNA is performed.55 IDUS is being increasingly used to better stage these cancers, especially portal vein or hepatic artery invasion or pancreatic involvement.
Gallbladder polyps are incidental findings on abdominal ultrasound. Polyps larger than 1 cm are more likely to be neoplastic. Previously it was difficult to differentiate neoplastic from nonneoplastic polyps, however, a group from Japan has had recent success using EUS morphology. A tiny echogenic spot or an aggregation of echogenic spots and multiple microcysts or a comet-tail artifact indicates a nonneoplastic polyp. Comparing EUS and ultrasound with surgical specimens, they found that EUS was 97% accurate versus 76% for ultrasound in differentiation of benign from neoplastic polyps.56-58
EUS is very accurate in tumor staging of gallbladder cancer. It can clearly delineate an inner hypoechoic and an outer hyperechoic layer in the wall of the gallbladder. Therefore EUS can accurately stage depth of invasion, as a recent study from Japan revealed, when EUS depth correlated well with histological depth.59 Nodal staging, even with FNA, is a challenge due to difficulty in visualization or positioning of node for FNA, along with a high false-negative rate. It is poor for distant metastasis and a CT scan is most useful in this evaluation.
Today, the liver is scanned routinely when staging for gastrointestinal and pulmonary malignancies with EUS. In 1999, a study was conducted in which 574 patients with suspected malignant gastrointestinal or pulmonary tumors undergoing upper EUS examinations also underwent EUS evaluation of the liver. Fourteen (2.4%) patients were found to have focal liver lesions on which FNA was performed. Fourteen of the 15 lesions sampled (1 patient had 2 liver lesions aspirated) were malignant. Interestingly, only 3 of the 14 patients (21%) had liver lesions depicted on prior CT scanning.60 Since then, several studies have confirmed the safety and significant clinical impact on patient management of EUS and EUS-guided FNA of liver lesions.61-64
Esophageal cancer involving the celiac lymph nodes is staged M1a and is managed by chemoradiation prior to esophagectomy, whereas earlier stages can be managed by surgery alone. In a study assessing the accuracy of EUS in detecting malignancy in celiac lymph nodes based on four criteria (>1 cm, round, homogeneous echo pattern, sharp borders), sensitivity was 83% with a 98% specificity.65 Accuracy is increased to 98% with the addition of FNA.66 EUS-guided FNA is superior to CT scanning in detecting malignant celiac lymph nodes. A recent study indicated that CT detected only 30% of malignant nodes diagnosed by EUS-guided FNA and that EUS-guided FNA significantly affected subsequent patient management.67
The left adrenal gland is visualized on EUS as “seagull-shaped”; the right is not usually seen. Metastatic cancer to the adrenal gland carries a grave prognosis and precludes curative surgical resection. Eloubeidi and colleagues68 found that EUS-guided FNA of the left adrenal gland was safe and confirms or excludes malignant adrenal involvement in patients with thoracic or gastrointestinal malignancies.
EUS can accurately assess mediastinal lymphadenopathy. Lung cancer is the most common cause of cancer-related death in the West and about 80% of cases are due to non–small cell lung cancer (NSCLC). Unfortunately, nearly 50% of patients harbor mediastinal disease at presentation. Direct mediastinal invasion or spread to contralateral lymph nodes (N3) is staged as IIIB disease; it is unresectable and treated with chemoradiation. Metastasis to ipsilateral and subcarinal nodes (N2) is classified as stage IIIA disease. Management is controversial. Most centers treat with chemoradiation only, but some centers offer surgery after adjuvant therapy. EUS can visualize posterior lymph node stations only (5 and 7), as air in the trachea prevents visualization of the anterior stations (2 and 4; Figure 10). Mediastinoscopy can stage anterior lymphadenopathy, but is associated with a complication rate of 2–3% and requires general anesthesia and a longer hospital stay. It cannot evaluate posterior lymphadenopathy. Nevertheless, both investigations should be regarded as complementary. EUS is highly accurate with a sensitivity of approximately 93% and a specificity of nearly 100%.69 EUS has been found to be superior to CT and PET scanning.70 In a recent large study EUS-FNA detected mediastinal disease in 10 of 24 patients (42%) without mediastinal adenopathy on CT.71
Another study by Wallace and associates72 in patients with NSCLC with nodes less than 1 cm by CT revealed advanced disease in 25% of patients by EUS-guided FNA. Transbronchial node aspiration (TBNA) performed during bronchoscopy is usually a blind procedure limited to pretracheal, paratracheal, and subcarinal areas, with yield varying from 20% to 70%. Recently, Krasnik and associates73 reported promising experience with a prototype realtime endobronchial ultrasound (EBUS)-TBNA bronchoscope. The same investigators in a later study combined transesophageal EUS-guided FNA and EBUS-TBNA biopsy in the evaluation of mediastinal masses. They were found to be complementary and the accuracy for diagnosis of mediastinal cancer was 100%. The investigators concluded that the combined approach may be able to replace more invasive methods for evaluation of mediastinal cancer.74 Currently, EBUS is being performed at select centers in the United States and Europe. EUS is also very accurate in diagnosing other diseases affecting the mediastinum, such as sarcoidosis, histoplasmosis, tuberculosis, lymphoma, and metastasis from other primary tumors.75
EUS is a new endoscopic tool that has revolutionized how gastrointestinal diseases are managed, particularly gastrointestinal tumors. It has numerous diagnostic and therapeutic indications, which continue to expand. There is a steep learning curve and a need for more gastroenterologists to be trained in this advanced procedure. It usually takes a dedicated year of training to become reasonably skilled in the technique. Parada and colleagues,76 in a resource utilization projection study, concluded that current EUS resources are not sufficient to meet the demand in the United States. If more gastrointestinal programs offered advanced fellowships in EUS, this potential problem could be averted. Higher reimbursement for the procedure could make the sacrifice of an additional year of fellowship worthwhile. Reduced equipment costs (currently between $300,000 and $350,000) could also help mainstream EUS into community practices. The ultimate goal is to make EUS easily accessible to all gastroenterologists, so that patients do not have to travel great distances and the most effective management options are available at home.