Subepithelial lesions

Introduction

The term subepithelial lesion (SEL) is used by endoscopists to describe any bulge covered with normal mucosa, usually found incidentally during gastrointestinal (GI) endoscopy or barium contrast radiography. Actually, this lesion could be either an intramural mass or an impression caused by extramural structures. Previously the term “submucosal tumor” was used but the term “subepithelial lesion” is frequently used now because intramural lesions may arise from any layer of GI wall underneath epithelium and the lesion might not be a tumor. The precise incidence of gastric SEL found during routine endoscopy is not known, but the detection rate seems increasing, especially with regard to small lesions, due to the advances in endoscopic technology and enhanced endoscopists’ attention.

To characterize the cause of protrusion, some noninvasive imaging methods such as transabdominal ultrasonography, computed tomography (CT), and magnetic resonance imaging (MRI) have been used, but they are often insufficient. With endoscopic ultrasonography (EUS), however, clinician can visualize the structure of gut wall layers clearly. Thus, EUS can not only differentiate SELs from extramural structures but also identify the layers of origin and endosonographic characteristics of intraluminal lesions. EUS is now accepted as the modality of choice for visualization of SEL with high precision.

The differential diagnosis of SEL includes a wide variety of benign and malignant subepithelial neoplasms, as well as nonneoplastic lesions ( Video 10.1 ). To evaluate SEL, the transition zone (the area where the tumor arises from normal gut wall layers) should be examined carefully to determine the layer of origin. Next, the size and echo pattern of the tumor, such as echogenicity, the smoothness of the border, internal features, and vascularity, should be observed. In addition, the relationship with other adjacent organs and the presence of adenopathy nearby provides valuable information. From the information gathered, an educated guess on the SEL for the differential diagnosis can be made with reasonable accuracy ( Table 10.1 ). However, reported accuracy of imaging diagnosis with EUS in predicting the pathology of SEL varies widely, and EUS-guided tissue acquisition can increase the diagnostic accuracy.

Diagnostic information on the SELs, including the origin of the wall layer provided by EUS, also helps in deciding whether a lesion should be removed or followed in situ. Lesions confined to the mucosal or submucosal layers can be safely removed endoscopically. Surgical resection, if needed, is generally recommended for lesions located in the muscularis propria, although advances in endoscopic techniques such as endoscopic submucosal dissection (ESD), endoscopic full-thickness resection (EFTR), or submucosal tunneling endoscopic resection (STER) have made it possible for these lesions to be removed by experienced clinicians with minimal risk to the patient. EUS can be used for following up after resection.

Comparison of accuracy between EUS and other imaging modalities

Compared with endoscopy, barium contrast radiography, transabdominal ultrasonography, CT, and MRI, EUS has a higher accuracy in detecting and assessing the size and location of SEL. When viewed endoscopically, the surface of SEL is usually smooth and has a color similar to that of the surrounding mucosa, without ulceration or erosion. Sometimes these lesions show a slight color change and certain morphologic characteristics, but it is often impossible to differentiate them by endoscopy alone. Ultrasonography may provide diagnostic information for large SEL. In a study of patients with endosonographically diagnosed gastric SEL, 94% of tumors (less than 3 cm in size) were visualized by transabdominal ultrasonography after the stomach was filled with echoic cellulose-based gastric contrast agents. Ultrasonography can also provide useful information on perigastric structures. CT may be used to evaluate large SEL especially when it is malignant and metastasis is suspected. However, a study pointed out that small SEL previously identified by EUS were not observed with preoperative multidetector CT. Accuracy of CT in differential diagnosis of SELs was 50.9%.

In addition to detection, only EUS can establish the precise location of the lesion within the GI wall and provide information on the sonographic characteristics of the SEL. The narrow differential diagnosis of SEL afforded by the use of EUS enables appropriate management decision making. Based on EUS, the clinician can decide between observation with re-examination, in patients with suspected benign lesions, and resection, when the lesion is likely to be malignant.

In the differentiation between SEL and extraluminal compression, EUS also demonstrates higher accuracy than endoscopy, ultrasonography, and CT. In a multicenter study, endoscopy was able to differentiate SEL from extraluminal compressions with sensitivity and specificity of 87% and 29%, respectively. In another study, of the 70 endoscopically suspected SEL, EUS detected 56 SEL and 14 external compression. CT detected only 39 cases of SEL. CT undetected cases included external compression and small-sized SEL.

Extramural lesions

Because EUS is able to visualize the gut wall layers in detail, it can readily differentiate the intramural and extramural nature of subepithelial mass-like lesions. When EUS demonstrates the integrity of all gut wall layers between the gut lumen and the lesion, it is safe to say that the lesion is an impression caused by an extramural structure.

The extramural structures that compress the gut wall are more likely to be adjacent normal structures, although it could occasionally be due to pathological lesions ( Table 10.2 ). A study revealed that when EUS evaluation was done for the patients with suspected extraluminal compression or SELs during endoscopy, 66.4% of them were proven to be extraluminal compression. It is worth noting that only 11% were due to pathologic lesions and others were related with adjacent normal organs or vessels.

A normal spleen usually makes an impression in the gastric fundus and upper body, and the gallbladder compresses the gastric antrum. Transient gastric impression is often caused by bowel loops. Other causes of gastric impression include vessels in the splenic hilum, the pancreatic tail, and the left lobe of the liver. Abnormal structures such as pancreatic pseudocysts, splenic artery aneurysm, aortic aneurysm, cystic tumor of the pancreas or liver ( Fig. 10.1 ), colonic tumors, and lymphoma may also produce endoscopically visible impressions on the gastric wall. Adjacent structures, such as the aortic arch and vertebrae, can also press on the esophagus. Other potential causes of esophageal impression are vascular anomalies, such as a right descending aortic arch, anomalous branches of the aortic arch, aneurysm, and left atrial dilation. Enlarged mediastinal lymph nodes or mediastinal tumors, lung cancer, and lymphomas are also known to compress the esophagus.

When using EUS, the suspected area of gastric impression should be observed by the two-step method. First, at a low frequency of 7.5 MHz, the examiner should survey the gross relationship between the extramural structure and the gut wall. Then, at a higher frequency of 12 MHz, the outer hyperechoic serosal layer should be observed carefully to determine whether it is intact or disrupted. This method allows reliable differentiation between gut wall impression and gut wall infiltration caused by an extraluminal tumor. For examination of small lesions, a high-frequency catheter ultrasound probe is technically easier to use than is a conventional echoendoscope. In the esophagus, the endosonographer may encounter difficulties in this evaluation owing to interference from the air-filled bronchial system.

Evaluation of subepithelial lesions

Gastrointestinal stromal tumor

Gastrointestinal stromal tumors (GISTs) are some of the most common mesenchymal tumors in the GI tract, and they are also the most commonly identified intramural subepithelial mass in the upper GI tract. Previously, these tumors were classified as GI smooth muscle tumors, such as leiomyomas and leiomyosarcomas, owing to histologic findings of circular palisades of spindle cells with prominent nuclei and apparent origin in the muscularis propria layer of the gut wall. However, with the development of newer molecular markers and an improved understanding of the biologic behavior of these tumors, GISTs are now classified as a distinct but heterogeneous group of mesenchymal tumors with varying differentiation. Interstitial cells of Cajal, also known as pacemaker cells of the GI tract, are now believed to be the precursor of GISTs that typically expresses c-KIT proto-oncogene, a transmembrane tyrosine kinase receptor. With immunohistochemical staining techniques, most GISTs stain positive for CD117, epitope of KIT protein, and, sometimes, CD34 but negative for desmin. Leiomyomas express smooth muscle actin and desmin, however, and schwannomas produce S-100 protein and neuron-specific enolase.

According to recent classification, approximately 70% to 90% of SEL from the fourth layer of stomach are GISTs, and approximately 10% to 30% of GISTs are malignant. Leiomyomas are the most common mesenchymal tumors in the esophagus, but they rarely occur in the stomach and small bowel. In contrast, GISTs are rare in the esophagus and are more common in the stomach (60% to 70%) and small bowel (20% to 25%).

The most common symptoms associated with GISTs are vague abdominal discomfort and pain, but most lesions are small (<2 cm) and asymptomatic. Larger lesions (>2 cm) may be ulcerated on top of the mass, and patients may present with bleeding or anemia. Occasionally, GISTs cause intestinal obstruction.

In defining the prognosis of patients with GIST, it has been recommended that a “grading as to the risk of aggressive behavior” be used instead of the term “benign.” This means that no GIST can be definitively labeled as benign, and all are considered to have some malignant potential. Pathologists classify GISTs as “very low risk,” “low risk,” “intermediate risk,” and “high risk” according to the size of the mass and the mitotic count of the resected specimen.

Endosonographically, a GIST is typically a well-circumscribed, hypoechoic, relatively homogeneous mass that can arise from either the second hypoechoic layer (muscularis mucosa) or, more frequently, the fourth hypoechoic layer (muscularis propria) ( Fig. 10.2 ). In contrast, leiomyomas ( Fig. 10.3 ) arise from muscularis mucosa more frequently than do GISTs. The images of GISTs, leiomyomas, and schwannomas are seen as relatively homogeneous hypoechoic masses under EUS and cannot definitely be differentiated unless performing special immunohistochemical tissue staining on resected tissue. One study suggested that GISTs have a marginal hypoechoic halo and relatively higher echogenicity compared with the adjacent muscular layer. Another study added inhomogeneity and hyperechoic spots to the foregoing features, and the presence of at least two of these four features predicted GISTs with 89.1% sensitivity and 85.7% specificity.

Lately some researchers tried to adopt artificial intelligence (AI) in differentiating GIST from other SEL. In one study, images of 173 SEL cases including 112 GIST, 43 leiomyoma, 7 schwannoma, 7 aberrant pancreas, 2 lipoma, and 2 inflammatory lesions were used for AI training. After training, diagnostic accuracy of AI was 86.3 % for SELs less than 20 mm and 90.0% for SELs larger than 20 mm which were higher than the accuracy of experts (73.3% and 53.3%, respectively). Another AI study tried to differentiate GIST from leiomyoma and AI showed a diagnostic accuracy of 96.3%. But these studies have limitations of retrospective analysis with selection bias using high-quality images. One multicenter prospective diagnostic study reported that diagnostic accuracy of endosonographers was improved from 69.7% to 78.8% with AI assistance. In the future, AI maybe a helpful ancillary tool during EUS procedure in differentiating GIST.

There have been several studies to predict potential malignancy of GIST based on the EUS characteristics of the lesion, but none of them obtained completely satisfactory results. In addition to size and mucosal ulcer, other EUS characteristics were considered as possible predictive factors, but size was the only consistently definitive predictive factor. ^, EUS features mentioned by the researchers were distorted shape, lobulation, irregular border, increased echogenicity in comparison with the surrounding muscle echo, inhomogeneity, hyperechoic spots, anechoic area, marginal halo, and extraluminal growing pattern. In one study, internal hypoechoic feature was suggested as a predictive marker of the tumor enlargement potential. When malignant changes occur, GISTs commonly show heterogeneous echo texture with hyperechoic deposits or anechoic necrotic zones inside large tumors ( Fig. 10.4 ). In one report, EUS findings of tumor size greater than 3 cm, nodular, heterogeneous, anechoic spaces, and ulceration were strong indicators of malignancy. A multicenter study reported that malignancy or indeterminate GIST status correlated with the presence of ulceration, tumor size larger than 3 cm, irregular margins, and gastric location but not with hyperechoic or hypoechoic internal foci. Nonetheless, a lack of defined risk factors could not exclude a malignant potential.

Recently, contrast-enhanced harmonic EUS (CEH-EUS) has been introduced. Using ultrasound contrast medium that includes microbubbles containing carbon dioxide gas with a lipophilic shell, the enhancement of echo levels in target tissues can be obtained because these bubbles are oscillated by the acoustic stimuli. The image of GIST is hyperenhanced after infusion of ultrasound contrast, in consequence, CEH-EUS signal intensity of GIST is higher than other benign lesions. SonoVue (Bracco SpA., Milan, Italy) or Sonazoid (Daiichi-Sankyo, Tokyo, Japan) are frequently used contrast agents. Studies that showed hyperenhancement had 73% to 100% sensitivity, 60% to 100% specificity, and 82% to 100% accuracy in distinguishing GIST from benign SEL ( Table 10.3 ). It is suggested that hyperenhancement and avascular areas in the center of the lesion are shown in GISTs but not in leiomyomas ( Fig. 10.5 ). In addition, prediction of malignant risk of GIST was possible with CEH-EUS by identifying intratumoral irregular vessels or nonenhancing spots with 53% to 92% accuracy ( Table 10.4 ). ^{, ,}

Use of EUS elastography for the differential diagnosis of gastric SELs has recently been suggested. Elastography provides images and measurements related to tissue stiffness. Elasticity of SEL can be measured by EUS elastrography. On EUS elastography, benign SEL shows an intermediate stiffness with homogeneous strain pattern, but malignant SEL shows a heterogeneously stiff pattern. It might be helpful for differentiating GIS Ts from other SELs because GISTs are usually harder than other benign SELs including leiomyoma. ^,

EUS-guided tissue acquisition can be performed for immunohistochemical examination to achieve definite diagnosis of GIST. Previously EUS-guided fine-needle aspiration (EUS FNA) and EUS-guided Tru-Cut biopsy (EUS TCB) were performed, but their performance was not good enough. A major drawback of EUS FNA is its inability to differentiate benign from malignant GISTs with absolute certainty. However, staining for ki-67 (MIB-1), a marker of cell proliferation, may enable the discrimination of benign from malignant GIST with EUS FNA. ^, Although higher rate of histologic specimens can be harvested using a 19-gauge Tru-Cut needle (Quick-Core, Cook-Medical, Bloomington, IN, USA), there are technical limitations. Recently, EUS-guided fine-needle biopsy (EUS FNB) with various newly designed biopsy needle is adopted. One study compared FNA needle (Echotip Ultra, Cook Endoscopy, Winston-Salem, NC, and Expect, BSCI, Marlborough, Mass) and FNB needle (SharkCore, Medtronics, Dublin, Ireland) for sampling of suspected GIST. Diagnosis with immunohistochemical staining was possible in 52.7% (EUS FNA) and 86.7% (EUS FNB).

Because small (<1 cm), asymptomatic mesenchymal tumors are rarely malignant, a policy of close follow-up with EUS may be justified, although an optimal surveillance strategy has not yet been established. Excision is advised when growth of the lesion, a change in the echo pattern, or necrosis is noted during annual follow-up with EUS. Surgical treatment is indicated for lesions greater than 3 cm in diameter with features suggestive of malignancy. For lesions between 1 and 3 cm, EUS-guided tissue acquisition can be recommended, or endoscopic resection can be chosen as a definite diagnostic and therapeutic tool with some risk of bleeding and perforation (2% to 3%, in specialized centers). When the lesion is confirmed to be a GIST, the risk of malignant transformation needs to be discussed with the patient; more careful follow-up or early resection should be considered.