The technical aspects of EUS-FNA are covered in detail in Chap. 23.

Several studies have demonstrated that EUS-FNA is a safe and accurate means of diagnosing subepithelial lesions of the gastrointestinal tract, particularly GISTs, with overall accuracy rates ranging from 67 to 98 % (Video 28.1) [2, 9, 22–30]. In the largest relevant study published to date, comprising 141 patients with gastric subepithelial lesions , the overall accuracy rate of EUS-FNA was 96 % based on criterion standard (surgical histopathologic results, or follow-up course for inoperable cases) [23]. However, diagnostic yield of EUS-FNA may be somewhat limited with EUS-FNA being diagnostic in 43–68 % of cases [31].

Factors that may enhance the diagnostic yield of EUS-FNA include the presence of an on-site cytopathologist, higher number of needle passes (five are recommended), and availability of immunohistochemical staining. Needle diameter has not been definitively shown to significantly impact the diagnostic accuracy of EUS-FNA for subepithelial lesions [32, 33], but 25-gauge needles may more easily puncture small, mobile lesions, as well as those within or adjacent to the duodenum when the scope tip may be acutely angulated.

In cases where tissue architectural details and immunohistochemical staining are required, obtaining a core-tissue specimen via EUS-FNB may be advantageous [34]. Another potential advantage of obtaining tissue cores is that specimen adequacy can be determined by the endoscopist, whereas FNA samples require an on-site cytopathologist. Combining EUS-FNA with FNB may be superior to either tissue sampling technique alone, [35] although this approach has not been extensively studied in patients with subepithelial lesions.

In select cases, EMR or ESD of subepithelial lesions may be performed to simultaneously obtain a histologic diagnosis as well as provide definitive treatment. This approach may be considered for situations in which FNA or FNB is likely to be low yield (e.g., very small lesions, suspected symptomatic pancreatic rest) or when previous stacked biopsies were diagnostic for a lesion that warrants resection (e.g., carcinoid tumor, granular cell tumor). Although associated with an increased risk for complications , endoscopic resection of lesions arising from the submucosa and even muscularis propria is increasingly performed and has a high diagnostic yield (87–94 %) [10, 16, 36–41]. It is necessary to identify the layer of origin with EUS before attempting resection because the risks are directly related to the depth of the tumor. Traditional saline-assisted polypectomy and cap-assisted EMR may be used to resect lesions. A relatively simple and elegant way of resecting small lesions that arise from the deep mucosa or submucosa (without sonographic evidence of involvement of the muscularis propria) is endoscopic band ligation with snare polypectomy. This technique is frequently employed for the resection of rectal carcinoids smaller than 1 cm in diameter, and has been shown to be superior to conventional polypectomy in terms of achieving complete resection with negative margins [42, 43]. Band ligation with or without electrosurgical resection has also been employed as a promisingly safe and effective method of treating small subepithelial lesions arising from the muscularis propria, including GISTs [37, 44–46]. In the so-called “ligate and let-go” technique, snare resection is not performed at the time of band ligation, thereby avoiding the risks of bleeding and perforation . Rather, the lesion is allowed to undergo ischemic necrosis and spontaneously slough off over time. The long-term effectiveness of this technique as a treatment option remains to be shown and a downside to this technique is the lack of a complete specimen for histologic examination.

GISTs typically are round/oval, firm lesions with smooth contour and normal overlying mucosa, although ulceration may be present with larger tumors (Fig. 28.6). Endosonographically, GISTs are typically hypoechoic and most commonly originate from the fourth EUS layer (muscularis propria). Important features to assess by EUS include the size, regularity of the outer border, and presence of echogenic foci and cystic spaces. Large tumor size (> 3 cm) and irregular border are the most reliable predictors of malignant behavior; other less consistent predictors include heterogeneous echotexture, cystic spaces, extraluminal growth, and hypervascularity [53–57].

GISTs were originally considered smooth muscle tumors , but are now known to arise from the interstitial cells of Cajal, which are the pacemaker cells of the gastrointestinal tract. Histologically, the majority of GISTs are composed of spindle cells arranged in interlacing, short fascicles or in a storiform pattern of growth (Fig. 28.6). A smaller proportion of GISTs are composed of epithelioid cells or a mixed cellular composition. The hallmark immunohistochemical feature of GISTs that distinguishes them from other mesenchymal/spindle cell tumors is positive staining for CD117 (c-KIT), which is expressed in over 90 % of GISTs [58–60]. A novel marker known as DOG1 (discovered on gastrointestinal stromal tumors 1) is comparable to CD117 in terms of sensitivity and specificity, and may be especially useful in diagnosing cases of CD117-negative GISTs [61, 62]. Other markers that may be expressed include CD34 (60–80 %), and less commonly smooth muscle actin (SMA) and S100.[58]. While these markers are generally unhelpful in confirming a diagnosis of GIST, they are useful in the diagnosis or exclusion of other gastrointestinal mesenchymal tumors [63].

Patients with GISTs should ideally be managed by a multidisciplinary team with expertise in sarcomas or tumors of the gastrointestinal tract [64, 65]. Gastroenterologists, working in conjunction with pathologists, are usually responsible for establishing the diagnosis and facilitating the appropriate referrals. Surgeons and medical oncologists are primarily responsible for developing a comprehensive treatment plan based on the resectability of the primary tumor, the aggressiveness of the tumor (Table 28.3), and the extent of any possible metastases.