Applications of Endoscopic Ultrasonography

, , and , Harbor-UCLA Medical Center, Torrance, Calif.


Medscape General Medicine. 1999;1(3) 

In This Article

Applications of EUS

I. Diagnosis and Staging of Benign and Malignant Lesions Within and Adjacent to the Gastrointestinal Tract

EUS can detect lesions as small as 2-3 mm in size and is the best method for determining the five echo layers of the stomach. The five echo layers are histologically correlated with the mucosa [layer 1], muscularis mucosa [layer 2], submucosa [layer 3], muscularis propria [layer 4], and serosa or adventitia [layer 5] (see Figure 3).

Figure 3. Layers of the stomach, as depicted by Olympus radial scanner.

Gastric submucosal lesions. Submucosal lesions are one of the primary indications for EUS. This preoperative evaluative technique can accurately differentiate between extramural compression, vascular lesions, and solid tumors. EUS can also accurately detect the depth of tumor invasion or the wall layer from which the tumor arises. Before an attempt is made to biopsy a submucosal lesion, an EUS exam should be performed to rule out extramural compression or vascular lesions. Table 1 outlines characteristics of typical submucosal lesions of the stomach located within the stomach wall.[12]

Leiomyomas are the most common type of submucosal tumor found by EUS. They usually arise from the muscularis propria [fourth echo layer] (see Figure 4), but small tumors may arise from the second echo layer of the stomach. Leiomyomas are generally hypoechoic and homogeneous in appearance with regular borders. Recently, a group of endosonographers evaluated the most useful criteria in differentiating benign from malignant stromal tumors by EUS findings. Tumor size >4 cm, irregular borders, echogenic foci, and cystic spaces were all dependently associated with malignancy in these studies.[13] When images were reviewed by different endosonographers, and two of these features were present, the sensitivity of EUS in predicting malignancy was 80% to 100%; if none of these EUS features were present, malignancy occurred in only 0% to 11% of cases. However, large-core biopsy or surgical/endoscopic removal of the lesion is frequently still necessary to establish the diagnosis if any suspicious findings are present. EUS can be used in following benign lesions at intervals to evaluate for any evidence of progression.

Figure 4. Leiomyoma of the stomach, which connects to the muscle layer of the stomach wall.

Other submucosal lesions can also be identified by characteristic EUS appearance. Lipomas are generally homogeneous, hyperechoic, well demarcated, and located within the submucosal layer. Carcinoid tumors are mainly located in the second (muscularis mucosa) or third (submucosal) layer and have an intermediate echogenicity with well-demarcated borders. Cystic lesions tend to be anechoic and well demarcated. Ectopic pancreas occasionally can be found in the submucosal layer. The appearance of this entity is similar to pancreatic parenchyma, with fine scattered hyperechoic spots and, occasionally, duct-like structures with regular borders.

Large gastric folds. Large gastric folds can arise from benign as well as malignant conditions (Table 2) and therefore represent a challenge in diagnosis. Whereas gastritis and foveolar, and glandular hyperplasia can be easily assessed by mucosal biopsy, evaluation of diffuse types of gastric cancer (which may not involve the submucosa), lymphoma, and gastric varices may be difficult. If the fourth layer (muscle layer) is thickened on EUS, deep biopsies (including strip biopsies) should be performed. If biopsies are negative, surgical exploration is generally recommended to rule out gastric carcinoma.[14] Most patients with scirrhous carcinoma show thickened third and fourth wall layers. Gastric varices appear as hypoechoic vessels in the submucosa and in the region surrounding the stomach. EUS with Doppler can be used to confirm the vascular nature, but is usually not needed for diagnosis. These guidelines can be used in conjunction with endoscopy to evaluate and manage patients with large gastric folds.

Esophageal cancer. Endoscopic ultrasonography is the single best modality for staging esophageal cancer with respect to depth of tumor infiltration and extent of lymph node involvement using the TNM classification system (Figures 5-8). The accuracy of EUS for staging patients with esophageal carcinoma who subsequently underwent surgery for confirmation ranged from 59% to 92% (with a mean of 84% in 1154 patients); the accuracy for N staging ranged from 50% to 90% (with a mean of 77% in 1035 patients).[15] Endosonography is superior to CT in the T and N staging of esophageal cancer (Table 3).

Figure 5. TNM staging of esophageal carcinoma. M=metastasis; N=lymph node; T1-T4=depth of esophageal wall invasion.
Figure 6. Esophageal carcinoma stage T2. The tumor (TU) invades muscle layer but does not penetrate through wall. LN=lymph node.
Figure 7. Esophageal carcinoma (TU) penetrating through wall of esophagus (stage T3). LY-NO=lymph node.
Figure 8. Esophageal carcinoma (stage T4) involving aorta (partial encasement) with loss of fat plane (interface) between tumor and aorta.

At our institution (Harbor-UCLA), EUS is used for a stage-dependent treatment protocol in the following way:

  1. To differentiate mucosal from submucosal invasion in early (stage T1) esophageal cancer where local therapy (endoscopic strip biopsy or laser, or minimally invasive surgery) may be considered. Local treatments may be a treatment option only in cancers limited to the mucosal layer that have a low risk of lymph-node metastasis. According to findings by Roesch,[15] if tumor invasion into the submucosal layer is evident, lymph-node metastases are found in 30% to 70% of cases. CT of the esophagus usually cannot differentiate between mucosal and submucosal tumor invasion.

  2. To reliably diagnose stage T4 esophageal cancer (Figure 7) where tumor infiltrates into adjacent organs such as the aorta, heart, lung, and spine. These tumors cannot be cured if removed and pose significant surgical risks. The accuracy of EUS in determining T4-stage disease is superior to that of CT scan (91% vs 49%, respectively; N=80).[15]

  3. To stratify patients that have local lymph-node metastases (stage N1) for preoperative chemotherapy and radiation therapy (Figure 9). EUS is markedly superior to CT scan in diagnosing local lymph-node metastasis, with an accuracy of 75% versus 54%, respectively[15] (Table 3). Criteria for assessing malignant lymph-node invasion on EUS are: size greater than 1 cm, hypoechoic appearance, distinct borders, and round shape. If all of these features are present, malignant lymph-node invasion can be predicted with an accuracy of 80%[16] to 100%.[17] However, all 4 of these features are only present in one fourth of cases, thus making it difficult to predict lymph-node metastasis in the remaining three fourths. The recent development of EUS-FNA has helped to solve this problem. EUS-FNAs of lymph nodes have an accuracy of 80%.[18]

    Figure 9. Malignant lymph node (LN); >1 cm, homogeneous, round.

  4. To obtain reliable diagnosis of celiac axis lymph nodes that are staged as M1, thus precluding patients from curative surgery. However, high-grade strictures proximal to the gastroesophageal junction sometimes do not allow a full evaluation of this status by EUS.

EUS is not a sensitive tool in diagnosing tumor involvement of the trachea and bronchial tree, because these structures contain air. Bronchoscopy should be used for staging of proximal esophageal cancers. If esophageal cancers obstruct the lumen and the EUS endoscope cannot be advanced through, tumor staging is incomplete. Dilation of tumor strictures for endosonographic examination to allow full passage of the echoendoscope may lead to perforations (up to 24% in one report).[19] It therefore has been recommended that esophageal dilations and EUS should be performed on different days. Another study reported that if dilation were performed in sequential incremental manner prior to EUS, no increased risk for perforation was observed.[20] There are two potential solutions to this problem: (1) the use of blind endoscopic ultrasound probes which can be introduced over a guide wire, and (2) the application of small radial mini-probes which can be introduced through the working channel of a conventional endoscope.

In summary, EUS has markedly changed the treatment of esophageal cancers through the use of a stage-dependent treatment protocol.

Gastric cancer. TNM staging of gastric carcinomas is shown in Figure 10. The accuracy of EUS to correctly stage gastric cancer is 78% for T stage and 70% for N stage.[21] Compared with CT, EUS is markedly superior in T and N staging of disease (Table 3). In this setting, EUS appears useful in the following stage-dependent treatment protocols :

Figure 10. TNM staging of gastric cancer. T1-T4=depth of invasion of stomach wall; N=lymph node; M=metastasis.
  1. Differentiating between mucosal and submucosal infiltration in stage T1 cancers when local resection by endoscopy is considered. High-frequency (15 or 20 MHz) ultrasound probes may be especially useful in these cases.[22] Local resection by snare biopsy techniques may also be considered in this clinical setting.[23]

  2. Identifying resectable lesions (T1-T3). See Figures 11-13.

  3. Identifying unresectable lesions that show evidence of local organ infiltration (stage T4; Figure 14). The accuracy of EUS-determined stage-T4 disease is about 80%.[21] In our own series with 34 patients, T4 lesions were determined with a sensitivity of 77% and a specificity of 100% (Eysselein and colleagues, unpublished data, 1999).

  4. At Harbor-UCLA, patients with distant lymph-node metastasis (N2) are considered for preoperative chemotherapy. In a recent study, patients with locally advanced gastric cancer (Stage IIIA,B and IV) as determined by EUS benefited from preoperative neoadjuvant chemotherapy, which increased the rate of RO resections (histologically complete tumor removal) to 80%.[23] Twenty-five percent of patients showed no recurrence of tumor. Most of the trials that have evaluated the utility of neoadjuvant chemotherapy for gastric cancer did not use EUS to clearly identify tumor stage. A prospective, randomized, multicenter trial studying the efficacy of preoperative adjuvant therapy based on staging by EUS is needed to show whether this application will alter the survival for locally advanced gastric cancer.

Figure 11. Gastric cancer (TU) confined to mucosa (stage T1, arrows).
Figure 12. Gastric cancer (TU) involving muscle layer of stomach (stage T2), but not penetrating through serosa. PV=portal vein.
Figure 13. Gastric cancer (TU) penetrating through wall of stomach (stage T3, arrows). PV=portal vein; SV=splenic vein.
Figure 14. Gastric cancer (TU) invading pancreas (stage T4, arrows). LN=lymph node; PV=portal vein.

Pancreatic carcinoma. TNM staging of pancreatic carcinomas is shown in Figure 15. The high mortality associated with pancreatic cancer confirms the need for early diagnosis and accurate staging to identify those patients who are candidates for resection. EUS is highly accurate in determining the T stage (80%) and N stage (72%) of pancreatic carcinomas[24] and is superior to conventional CT in T and N staging of disease.[24]

Figure 15. TNM staging of pancreatic carcinoma. T1-T4=local tumor extension; N=lymph node; M=metastasis.

In a study by Roesch and colleagues,[25] EUS had a 99% sensitivity in detecting pancreatic carcinomas, compared with 77% for CT and 90% for ERCP. EUS appears especially useful in the detection of small tumors (<3 cm; Figure 16), with a sensitivity of 100% versus 55% for CT and 90% for ERCP. Newer imaging modalities such as helical ("spiral") CTs appear superior to conventional CT scan. In a recent study by Bluemke and coworkers,[26] pancreatic carcinoma was directly visualized by spiral CT as a tumor in 89% of cases. In this study, only single-phase CT scans were used. Scanning was performed after early arterial phase of pancreatic enhancement, assessing mainly the portal phase of liver enhancement. Using dual phase (arterial and portal phase) helical CT, Ichikawa and associates[27] reported a tumor detection rate of 76%. All lesions >4 cm were detected by dual-phase helical CT; however, only 50% of the lesions <2 m were detected by this method.

Figure 16. Small (1.5 cm; stage T1) pancreatic carcinoma (TU). PD=pancreatic duct.

A recent study[28] reported 100% and 92% sensitivity for EUS and dual-phase helical CT for detection of pancreatic tumors, respectively. Sensitivity for small lesions (</= 15 mm) was 67% by CT and 100% by EUS. The accuracy for staging of pancreatic tumors and predicting resectability was 93% and 90%, respectively, using both techniques. The accuracy was slightly lower for predicting resectability/nonresectability of pancreatic cancers using EUS (76%) compared with helical CT (86%), as recently reported by Howard and colleagues.[29] In the study by Ichikawa and associates,[27] dynamic MRI appeared superior to dual-phase helical CT for the detection of small pancreatic lesions (<4 cm), with a detection rate of 83%. Further determination of the relative utility of dynamic MRI versus EUS for evaluation of pancreatic lesions is ongoing.

Assessment of vascular invasion is probably the most important parameter for determining resectability in pancreatic cancer. In a study by Roesch and colleagues,[30] EUS had a 95% accuracy for determination of portal vein involvement, versus 85% for angiography and 75% for conventional CT.

A study by Brugge and coworkers[31] compared in detail several criteria for evaluating vascular involvement using EUS and angiography. Criteria for identifying tumor invasion of major vessels were proximity of the tumor to the vessel, loss of interface between vessel and tumor, and irregularity of the vessel walls. These investigators demonstrated that the EUS finding of an irregular vessel wall accurately predicted vascular invasion and therefore nonresectability in all cases (100% specificity), compared with 94% specificity by angiogram using the same parameter. However, the finding of an irregular vessel wall was present in only 47% of cases with vascular involvement proven on surgery. Other criteria, such as loss of tumor-vessel interface or proximity of the tumor to the vessel, had a higher sensitivity (78% and 100%, respectively). However, false-positive findings occurred, thus decreasing the specificity to 79% and 44%, respectively.

In our experience, the most difficult aspect in the evaluation of resectability is tumor involvement of the superior mesenteric vein. Large tumors situated in front of this vessel diminish the penetration of the ultrasound beam. Brugge and associates[31] reported an EUS sensitivity of only 17% for determination of tumor invasion of the superior mesenteric vein using wall irregularity as the sole criterion. This result was comparable to that found on angiography (sensitivity of 20%). At our institution, an MRI angiogram is performed in cases where tumor involvement of the superior mesenteric vein is suspected by EUS, thereby replacing the need for conventional angiogram.

Figure 17 depicts an algorithm for staging pancreatic carcinomas as developed at Harbor-UCLA Medical Center. When pancreatic cancer is clinically suspected, a spiral CT (preferably with arterial and portal venous phase) is first performed. If the CT scan is indeterminate for tumor detection or resectability, EUS is performed. If the patient appears to have a resectable lesion by EUS, a staging laparoscopy is performed immediately prior to surgery. If the patient is deemed unresectable by EUS, MRI angiogram is performed to confirm tumor invasion of major vessels. Only patients who are thought to have an unresectable tumor undergo EUS-FNA (Figure 18) for tissue diagnosis of pancreatic carcinoma. EUS-FNA is also used to exclude lymphomas and to rule out infectious lesions (such as TB or Mycobacterium avium complex [MAC]). If chronic pancreatitis is present, and a tumor cannot be ruled out by EUS, EUS-FNA is performed and repeated within 3 months. At our institution, the accuracy for detecting pancreatic carcinomas by EUS-FNA is 95%, with a specificity of 100%. These results are in agreement with the literature (85% to 95%)[32,33] and support the usefulness of EUS in the work-up of clinically suspected pancreatic cancer.

Figure 17. Algorithm for work-up of pancreatic cancer at Harbor-UCLA Medical Center.
Figure 18. Fine-needle aspiration (FNA) of suspected pancreatic carcinoma. Hyperechoic (white) needle tip is marked by long arrow; arrowheads delineate tumor. TU=tumor.

EUS is one of the best methods to diagnose and stage ampullary/periampullary tumors, with accuracy of 80% for the T stage and 72% for the N stage.[34] EUS is the only method available for accurate local staging of papillary cancer. Cancers of the papilla can be treated by local resection, which at times can prevent the need for removal of the pancreas and the duodenum.

Pancreatic endocrine tumors. Endosonography, in conjunction with somatostatin receptor scintigraphy,[35] appears to be the most effective method for detection of pancreatic endocrine tumors (Figure 19). In one large study,[36] EUS had an accuracy of 92% (vs 25% by CT) in detecting these lesions. Interesting to note is that 82% of tumors not visualized by CT were found by EUS. EUS is now routinely used at our institution for detecting pancreatic endocrine tumors.

Figure 19. Small (15-mm) insulinoma in pancreas; arrows indicate tumor (dual-phase helical CT, MRI negative). SV=splenic vein.

Rectal cancer. EUS has gained wide acceptance for the preoperative staging of rectal carcinomas, with an accuracy of 85% for T staging and 79% for N staging (N=1344).[37] EUS is far superior to CT in terms of accuracy of T and N staging of disease (Table 3). EUS is essential for evaluation of patients with locally resectable rectal cancers (T-1 and 0) and for determination of the need for preoperative chemotherapy and radiation in patients with local lymph-node metastasis (N-1 and more advanced disease).

Chronic pancreatitis. EUS is as sensitive as ERCP in diagnosing chronic pancreatitis.[38] EUS detects not only changes in the pancreatic duct and its side branches, but also changes in the parenchyma, such as echogenic foci, small cysts, lobulation, and echogenic strands.[38,39,40] Important to note is that this technique does not lead to an increased risk of pancreatitis, which occurs as a complication of ERCP in approximately 5% of cases.

Choledocholithiasis. EUS has shown utility in diagnosing small stones of the common bile duct (CBD); the sensitivity and specificity of EUS in this setting reach those of ERCP. In a study by Sugiyama and colleagues,[41] EUS detected all stones >1 cm as confirmed by ERCP. EUS is especially useful for detection of small stones (3-5 mm), with a sensitivity of 91%--which is markedly higher than that of transabdominal ultrasonography (27%) and abdominal CT (27%). EUS is especially useful in the setting of clinically suspected gallstone pancreatitis, when laboratory tests (AST [aspartate transaminase], ALT [alanine transaminase], alkaline phosphatase, bilirubin) are only slightly or moderately elevated. In this situation, EUS can accurately diagnose or rule out CBD stones and avoid the risk of pancreatitis associated with an ERCP. A new bile duct imaging technique is emerging with magnetic resonance cholangiopancreatography (MRCP), which has a sensitivity of 81% in detecting CBD stones.[42] However, small stones near the papilla may be easily missed, thus limiting the use of MRCP (Eysselein and coworkers, 1999, unpublished results). MRCP may rival the use of EUS in the future and may be especially useful in patients who cannot tolerate endoscopy.

II. Interventional EUS

Drainage of pancreatic pseudocysts. EUS is currently used at several institutions,[43,44,45,46] including Harbor-UCLA Medical Center,[47] to evaluate the utility of endoscopic pancreatic pseudocyst drainage for treatment of chronic pancreatitis. Pseudocyst drainage is usually performed surgically through the stomach or duodenal wall. EUS can determine the distance between the wall of the stomach/duodenum and the cyst. A distance of more than 10 mm, however, usually prohibits an endoscopic approach.

EUS is also used to assess intervening vessels that would preclude pseudocyst drainage by endoscopy. The port of entrance into the pseudocyst is marked by EUS guidance, allowing safer endoscopic drainage with placement of stents.

Celiac axis blockade for pain management. EUS has also been used for pain management in pancreatic cancer and chronic pancreatitis to direct celiac axis blockade.[48,49] A recent study by Gress and colleagues[50] showed that EUS-guided celiac block provided more persistent pain relief than CT-guided celiac block. In this study, a local anesthetic (10 mL bupivacaine, 0.75%) and a glucocorticoid (40 mg triamcinolone in 3 mL) were injected on both sides of the celiac plexus region using EUS-FNA. However, further studies are needed to confirm the superiority of EUS-guided celiac axis blockade over the CT-guided approach.

III. Evaluation of Vascular Disease

EUS can accurately diagnose aneurysms of the aortic arch (Figure 20) and the descending aorta as well as vascular lesions such as thrombosis in the portal vein system.[51] Endosonography is also helpful in evaluating splenic vein thrombosis when other methods (such as transabdominal ultrasound and CT) are inconclusive.

Figure 20. Aneurysm of aortic arch with local wall thrombosis on CT scan. Arrowheads indicate aneurysm.

IV. Therapeutic EUS

Studies are currently under way to evaluate the utility of EUS-FNA in the local treatment of advanced pancreatic cancers to direct delivery of immunomodulatory therapy or tumor suppressor genes via recombinant viral vectors. These modalities are still experimental and their clinical applications have not yet been determined.