Standard endoscopy provides imaging of subepithelial lesions, whereas EUS can reveal the entire tumor and provide information regarding the size and origin of the tumor. EUS has become an important diagnostic tool in the evaluation of subepithelial lesions. High-frequency ultrasound imaging is capable of differentiating between intramural tumors, intramural vascular lesions, and tumors with extramural compression. EUS also provides valuable information about tumor shape, size, layer in which the tumor is situated, tumor border, regional lymphadenopathy, echogenic pattern of the lesion, and local spread of the tumor. With EUS, GIST have classically a round or oval shape and a dark or hypoechoic appearance, but they are comparatively hyperechoic to the muscle layer, and they typically arise in the fourth layer, which represents the muscularis propria (Figs. 3 and 4) [13, 24].

EUS has become a preferable diagnostic method to further evaluate subepithelial lesions discovered by ultrasonography (USG), computed tomography (CT), magnetic resonance (MR), or esophagogastroduodenoscopy. In a prospective study of 150 patients who had a presumptive diagnosis of a submucosal lesion in the gastrointestinal tract, the sensitivity and specificity of EUS to differentiate extramural lesions and submucosal lesions were 92 % and 100 %, respectively [25]. Many studies have demonstrated the capability of EUS to differentiate GIST from other subepithelial tumors. Okai et al. attempted to differentiate the various gastric mesenchymal tumors via the imaging of EUS. They reported that a marginal halo and a relatively higher echogenicity suggested a GIST. However, a marginal hypoechoic halo was also detected in patients with schwannomas [26]. In a study of 181 consecutive patients with submucosal lesions in the upper gastrointestinal tract, the authors reported a sensitivity and specificity of 95 % and 72 %, respectively, for the diagnosis of GIST by EUS [27]. Kim et al. reported that four features in particular – relative echogenicity, homogeneity, echogenic foci, and marginal halo – conferred a sensitivity and specificity of 89.1 % and 85.7 %, respectively, for the diagnosis of a GIST [28].

EUS imaging is currently used to differentiate gastrointestinal submucosal lesions and is certainly more effective in this regard than standard endoscopy. Nevertheless, EUS imaging alone is not able to differentiate gastrointestinal subepithelial lesions with sufficient accuracy. EUS is also highly dependent on the skill and experience of the endoscopist and is subject to variation in image interpretation [13, 29].

Most GIST are benign, yet malignant characteristics are observed in approximately 20 % of gastric GIST and 40–50 % of small intestinal GIST. Assessment of the malignant potential of GIST is a challenge to clinicians. EUS features can help to distinguish the malignant potential of GIST. Many studies have sought to define the EUS features that predict the malignant behavior of GIST [16, 28]. Pari et al. reported that the following EUS features could be used to predict the malignant behavior of a GIST: tumor size >5 cm, irregular extraluminal border, local invasion, and a heterogenous appearance. In this study, all lesions were completely removed surgically, and thus a diagnosis of GIST was secured with histology [30]. Kim et al. suggested that except for the tumor size and the irregularity of the tumor border, most of the EUS features were not useful in establishing the malignant risk of a given GIST. In patients with a tumor size ≥35 mm, they showed a sensitivity and specificity of 92.3 % and 78.8 %, respectively, for the diagnosis of GIST [28]. In a French study, the EUS criteria for malignancy were tested in 56 surgically resected gastrointestinal lesions and demonstrated that the presence of an irregular extraluminal margin, cystic spaces, and malignant-appearing lymph nodes were predictive of malignancy. The presence of at least one of these criteria had 91 % sensitivity, 88 % specificity, and an 83 % positive predictive value for malignancy [31]. Okai et al. found that exogastric growth, cystic changes, and echogenic foci within the tumor were not related to malignant behavior; however, lobulation of the lesion surface was often seen by EUS in the malignant GIST [26]. Jeon et al. found that certain EUS findings, including irregular tumor borders, mucosal ulceration, nonoval shape, and tumor size >3 cm, were correlated with a risk of malignancy [32].

In summary, there have been several studies to differentiate benign and malignant GIST based on EUS features. Irregular tumor border and tumor sizes have been associated with a risk of malignancy in the majority of studies. Thus, these parameters should be considered as predictive factors of malignancy for GIST when detected by EUS. Other EUS features, including echogenic foci, heterogeneity, and cystic spaces, have proven less consistent than irregular border and diameter of tumor in their capability to predict the risk of malignancy in a GIST [13]. New EUS techniques for tissue evaluation, including contrast-enhanced EUS, real-time elastography, and digital image analysis may be helpful in the differential diagnosis of GIST. Some studies have reported the capability of digital image analysis in differentiating benign from malignant subepithelial lesions on EUS [21, 33, 34].

Although EUS features can provide considerable clues to the risk of malignancy for GIST, the accuracy of EUS imaging features is not sufficient. Thus, tissue samples are often used to increase the diagnostic accuracy of EUS. Preoperative tissue sampling may not be necessary for large tumors or symptomatic tumors that need surgery regardless of the pathological diagnosis. The options for tissue sampling from GIST include EUS-guided fine-needle aspiration (FNA), EUS-guided fine-needle biopsy (FNB), EUS-guided Tru-cut biopsy, jumbo forceps or bite-on bite biopsy, surgical excision, endoscopic submucosal resection (ESMR), endoscopic submucosal dissection (ESD), percutaneous biopsy, unroofing, and tunneling techniques. Percutaneous biopsy is not usually recommended due to the risk of tumor rupture and peritoneal spread [13, 21, 35, 36].

There are no published large studies to evaluate the safety of EUS surveillance in patients with GIST. Optimal timing of surveillance with EUS has not been established for GIST. Frequency of EUS evaluation should be adjusted depending on the clinical scenario of patients. There are slight differences between guidelines in patients with GIST. The NCCN recommends that very small gastric GIST should be evaluated by abdominal/pelvic CT with contrast and/or EUS-guided FNA. If the tumor has no high-risk features including irregular border, cystic spaces, ulceration, echogenic foci, and heterogeneity, endoscopic surveillance at 6- to 12-month intervals may be considered. But complete surgical resection patients with very small gastric GIST with high-risk features should be considered [52]. The European Society of Medical Oncology (ESMO) and the NCCN released guidelines about management of GIST. The Guideline of ESMO recommends resection for gastric GISTs with diameter >2 cm. NCCN recommends removal of all GIST with size 2 cm or larger. However, resection is recommended for all GIST of diameter ≥3 and <3 cm with concerning EUS features including heterogeneity, an irregular tumor border, echogenic foci, and presence of cystic appearance by the AGA [13, 35, 52, 53].

EUS is a good diagnostic method for the differential diagnosis of subepithelial tumors. Also it is useful in the choice of the appropriate treatment method for GIST. EUS precisely determines the size of lesion, layer of origin, and growth pattern of tumor (intraluminal or/and extraluminal). All of these features are important features to use in the endoscopic resection techniques. An experienced team should be available before endoscopic resection [13, 20].