Diagnosis of gastrointestinal subepithelial lesions

Gastrointestinal (GI) subepithelial lesion (SEL) is a general term referring to lesions that develop in the submucosal or muscular layer of the GI tract, with most lesions unexposed on the mucosal surface. However, even if lesions of mucosal origin can be considered an SEL, lesions originating from the lamina propria or muscularis mucosae with normal mucosal epithelium and SEL-like elevations can also be included. Thus, SEL includes gastrointestinal stromal tumors (GISTs), leiomyomas, schwannomas, SEL-like carcinomas, malignant lymphomas, neuroendocrine tumors, ectopic pancreas, lipomas, simple cysts, bronchogenic cysts, lymphangiomas, amyloidosis, and many other diseases [1,2,3,4,5,6,7].

Histological evaluation should be performed to diagnose these SELs. However, obtaining tissues from SELs is difficult with conventional endoscopic forceps biopsy, except for the so-called delle lesions with mucosal surface depressions, because these lesions are covered by normal epithelium. Conversely, endoscopic ultrasound (EUS) is useful for SEL imaging because it can infer a diagnosis based on the GI wall layers where the lesion and its internal echo patterns are located. A submucosal lesion with a uniform non-echoic pattern is diagnosed as a cyst or lymphangioma. A submucosal lesion with a uniform hyperechoic pattern is diagnosed as a lipoma. These SELs can be diagnosed using EUS imaging alone. However, SELs with a hypoechoic pattern in the muscular layer include different mesenchymal tumors, such as GISTs, leiomyomas, and schwannomas [1,2,3]. Moreover, these tumors cannot be differentiated with EUS imaging alone [8]. Since GISTs are malignant and leiomyomas/schwannomas are benign tumors, the differential diagnosis of “GIST or not” is important in determining the treatment strategy for SELs with hypoechoic patterns in the muscular layer. SEL-like carcinomas and malignant lymphomas invading from the submucosa to the muscular layer also present a hypoechoic pattern [6]. Therefore, hypoechoic SELs in the submucosal and muscular layers should be pathologically diagnosed using tissue samples (Fig. 1).

Fig. 1
figure 1

EUS-FNA of SEL-like gastric cancer. A 3–4 cm mass lesion like an SEL was found in the greater curvature of the middle gastric body. a EUS using a radial-type echoendoscope revealed gastric carcinoma that invaded the muscular or serosal layers. b However, repeated forceps biopsies from the mucosal side did not yield tumor cells. Therefore, EUS-FNA was performed using a forward-viewing echoendoscope. c The puncture was performed using a 22-G fork-tip needle. e/f Hematoxylin/eosin staining showed mixed findings of poorly differentiated adenocarcinoma, well-differentiated tubular adenocarcinoma, and papillary adenocarcinoma. The patient was diagnosed with gastric cancer. EUS-FNA endoscopic ultrasound-guided fine-needle aspiration, SEL subepithelial lesion, EUS endoscopic ultrasound

Full size image

SEL tissue sampling methods

Previously, bite-to-bite biopsy (BBB) with forceps was performed to obtain tissue samples from SELs, a simple method of performing multiple biopsies obtained from the same site perpendicularly from the mucosal surface to the deeper layers of SELs. However, Ji et al. reported a low accuracy (38%) and a high bleeding rate (14%) [9]. Facciorusso et al. also performed propensity score matching between BBB and EUS-guided fine-needle aspiration (EUS-FNA) in upper GI and rectal SELs (120 patients in both groups) and found that the specimen collection rate (BBB vs. EUS-FNA, 77.5% vs. 94.1%) and accuracy (BBB vs. EUS-FNA, 67.1% vs. 89.3%) were significantly higher for EUS-FNA, whereas adverse events (BBB vs. EUS-FNA, 30% vs. 6.6%) were significantly lower for EUS-FNA [10].

On the other hand, mucosal incision-assisted biopsy (MIAB) based on endoscopic submucosal dissection (ESD) has been recently performed in East Asia, especially in Japan [11,12,13,14,15]. MIAB is performed by making a mucosal incision and submucosal dissection to expose the SEL, and the tissue is obtained under direct observation using forceps or other instruments. Its accuracy was reported to be 77.8%–100%, whereas some other studies reported it to be ≥ 90% [11, 12, 15, 16]. Especially in small SELs of < 20 mm in diameter, an accuracy of 75%–100% has been reported [16]. However, due to the complexity of the procedure, MIAB is limited to countries and institutions accustomed to ESD.

In 1992, Vilmann et al. used EUS-FNA as a method of performing needle biopsy under real-time EUS observation [17], which has become well known worldwide. EUS-FNA was first mainly indicated for pancreatic masses but later became the standard method of tissue sampling for SELs. Compared to MIAB, EUS-FNA has the advantage of simplicity, except for the need for a dedicated echoendoscope. Therefore, EUS-FNA is the first-choice technique for obtaining pathological findings from SELs.

Indications and diagnostic performance of EUS-FNA for SELs

EUS-FNA is indicated for SELs of submucosal or muscular origin that cannot be diagnosed with forceps biopsy [18,19,20]. However, cysts, lymphangiomas, and lipomas, which can be identified only on EUS images, are excluded. In practice, GISTs should be distinguished from other mesenchymal tumors, and since GISTs more frequently develop in the stomach, EUS-FNA for SELs has been commonly reported in gastric SELs [18]. EUS-FNA of esophageal and duodenal SELs has also been performed; however, the frequency of GISTs is lower than that of gastric SELs. In particular, leiomyoma, a benign tumor, is more commonly observed in esophageal SELs (Fig. 2) [21]. EUS-FNA of colorectal SELs has been performed for local recurrence after rectal cancer surgery, GIST, and ectopic endometriosis of the rectum, although only a few studies reported difficulty of deep insertion of the echoendoscope and the number of indicated lesions was small [22,23,24].

Fig. 2
figure 2

EUS-FNA of esophageal leiomyoma. a The lumen of the lower thoracic esophagus was narrowed by an SEL. b EUS with a radial-type echoendoscope showed a hypoechoic mass surrounding the esophagus. c EUS-FNA was performed using a forward-viewing echoendoscope. e The puncture was performed using a 22-G fork-tip needle. f Hematoxylin/eosin staining showed clusters of spindle-shaped cells. g Tumor cells were positive for desmin staining; C-kit staining was negative, and a diagnosis of leiomyoma was made. h EUS-FNA endoscopic ultrasound-guided fine-needle aspiration, SEL subepithelial lesion

Full size image

The accuracy of EUS-FNA for SELs varied from 43.3 to 100% depending on the study [16, 25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43], and the specimen collection rate and accuracy were considered lower than those for pancreatic masses. Most benign diseases such as leiomyoma and ectopic pancreas are diagnosed based on EUS-FNA and follow-up results without resection. In 2016, a meta-analysis reported by Zhang et al. demonstrated a pooled accuracy of 59.9% in 17 articles for EUS-FNA of upper GI SELs [44]. However, these articles were older, published from 2004 to 2014. Therefore, some overlap was found, and only two articles described the results of fine-needle biopsy (FNB) needles, which will be discussed later; five articles described Tru-Cut needle biopsy (TCB) needles [34, 45,46,47,48], which are no longer available; and 14 articles described FNA needles. The pooled accuracy for FNA needles was only 56%.

EUS-FNA for SELs has been used to diagnose gastric GISTs, the most common target among SELs, requiring a sufficient number of specimens for immunohistochemical staining. However, unlike pancreatic masses and swollen lymph nodes, obtaining specimens using EUS-FNA for SELs is difficult because the puncture target is within the GI tract wall, the lesion may escape with the wall during puncture, the lesion is hard to puncture, and even if the puncture is possible, moving the needle within the lesion is difficult [3, 20].

EUS-guided needle

As an alternative to the FNA needle, the TCB needle was expectedly useful as a puncture needle for obtaining core tissues [34, 45,46,47,48]. The TCB needle has the same structure as the needle used for liver biopsy. The inner needle was used to puncture the lesion, and the outer needle was inserted to cut out the lesion. However, use of the TCB needle was terminated in 2011, and it is no longer available due to its thick and rigid 19-G needle and the fact that it can only be used for lesions larger than 3 cm due to its structure.

To collect high-quality and high-quantity specimens, a puncture needle with a more improved shape compared to the conventional FNA needle known as an “FNB needle” is currently used [1, 49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73]. FNB needles include side-fenestrated reverse-bevel-type core trap, Franseen, and fork-tip needles. First, core trap needles with side-fenestrated reverse bevels were developed. Unlike core trap needles, the Franseen and fork-tip needles have an improved tip shape and are referred to as “new FNB needles.”

Facciorusso et al. compared FNA and FNB needles for SELs in their meta-analysis [70] and extracted 10 articles [56,57,58,59,60,61,62,63,64,65], including six randomized controlled trials from 2014 to 2019. They found that FNB needles consisted of seven reverse-bevel needles, three Franseen needles, and two fork-tip needles. The analysis concluded that the specimen collection rate (FNA vs. FNB, 80.6 vs. 94.9%, odds ratio 2.54) and accuracy (FNA vs. FNB, 65% vs. 87.9%, odds ratio 4.10) of the FNB needle were higher than those of the FNA needle [70]. In addition, a meta-analysis of FNB needles by Tan et al. revealed sufficient results with a specimen collection rate of 98.8% and an accuracy of 85.7% based on 16 articles from 2015 to 2020 [71]. However, among FNB needles, differences in diagnostic performance were observed, and EUS-FNA for SELs using the new FNB needle (such as Franseen and fork-tip needles) has been reported to have a sufficient specimen collection rate of 74–100% and an accuracy of 88–92.3% [1, 66, 67, 72]. In particular, the fork-tip needle has high puncture performance and is expected to provide a smooth puncture of SELs with a sharp sensation similar to that of FNA needles. Takasumi et al. reported a specimen collection rate of 100% and an accuracy of 92.3% for EUS-FNA with a 22-G fork-tip needle in gastric SEL [1]. Yamashita et al. reported that the specimen collection rate with fork-tip needles was higher than that with Franseen needles in SELs of < 2 cm (Franseen vs. fork-tip, 74% vs. 96%) [72]. However, this study used fork-tip needles after using Franseen needles, which may have been related to an improved surgical technique. Moreover, when EUS-FNA is performed using an FNB needle, the 22-G needle is considered the most commonly used. However, Antonin et al. reported that when EUS-FNA of upper GI SELs with a 22-G FNB needle was technically difficult or the specimen was poor, switching to a 25-G FNB needle facilitated the diagnosis in 56.2% of patients [73].

In summary, the FNB needle has a higher specimen collection rate and accuracy than the FNA needle in EUS-FNA for SELs (Table 1). Although the FNA needle seems to have better puncture performance, the FNB needle should be selected for SELs where a puncture with the FNB needle seems feasible. The European Society of Gastrointestinal Endoscopy guidelines strongly recommend the use of the FNB needle for EUS-FNA with SELs of > 20 mm and weakly recommend the use of the FNB needle for SELs of < 20 mm [18].

Table 1 Comparison of EUS-FNA with FNA and FNB needles in gastrointestinal SEL
Full size table

Technological devices

The type of echoendoscope primarily used in EUS-FNA is a convex-type echoendoscope. However, the conventional type is an oblique-viewing (OV) echoendoscope, which is designed to facilitate puncture of pancreatic masses. Therefore, no problems were encountered in puncturing SELs located in the posterior wall or lesser curvature of the stomach; however, puncturing SELs located in the greater curvature or anterior wall of the stomach was difficult.

In contrast, forward-viewing (FV) echoendoscopes are reportedly useful [74] and have the advantage of allowing puncture with the scope in front of SELs [74,75,76,77,78]. Yamabe et al. developed a method for performing EUS-FNA while aspirating the SEL by attaching a soft hood to the tip of the FV echoendoscope [77]. The lesion could not escape when aspirated into the hood, with a specimen collection rate of 87.5% in eight SELs of ≤ 15 mm (Fig. 3). Moreover, in a prospective, randomized, crossover study of 41 patients, Matsuzaki et al. reported no significant difference in the diagnostic yield (FV vs. OV, 80 vs. 73.3%) and accuracy (FV vs. OV, 77.2 vs. 72.7%) [78]. However, the FV echoendoscope had a larger median tissue sample area and a shorter median procedure time than the OV echoendoscope in patients with GIST. The use of FV echoendoscopes is expected to improve the success rate of SEL puncture from the greater curvature to the anterior wall in the gastric fornix and gastric body and contribute to the diagnostic performance of EUS-FNA. However, the usefulness of FV echoendoscopes for SELs should be validated in a prospective study.

Fig. 3
figure 3

A forward-viewing echoendoscope. a transparent hood made of polyvinyl chloride was attached to the tip of the scope for puncture while aspirating the mucosal surface of the lesion. b This figure shows the insertion of the FNA needle into the forceps channel of a forward-viewing echoendoscope. FNA fine-needle aspiration

Full size image

Negative pressure is generally recommended for EUS-FNA specimen collection and is used in two methods: the standard syringe aspiration and the “slow-pull method,” in which a stylet is slowly pulled out to apply low negative pressure [79, 80]. The advantage of the slow-pull method is the possibility to collect specimens with few blood components even in lesions with high blood flow. In particular, combining the FNB needle and the slow-pull method is expectedly useful. Lee et al. reported no difference in accuracy (standard vs. slow-pull, 81.5 vs. 83.3%) and volume of specimens collected between the standard suction and slow-pull methods in EUS-FNA of SELs using the FNB needle [81]. The results showed no difference in the number of specimens collected between the standard and slow-pull methods.

However, SELs may be hard and sometimes require slightly higher negative pressure. The “wet suction method,” defined as the concept of maintaining negative pressure aspirated with a syringe to the lesion by filling the puncture needle with saline solution, is reportedly useful mainly in diagnosing pancreatic lesions using EUS-FNA [82,83,84,85]. This method is also expectedly useful for specimen collection in EUS-FNA of SELs. Pita et al. reported that EUS-FNA was performed in 87 SELs using the wet method, which could be performed up to immunostaining in 81% of the patients [86]. Takasumi et al. conducted a prospective, randomized, crossover study comparing the cellularity of specimens from upper GI SELs collected via conventional aspiration and wet suction, but they failed to demonstrate the superiority of the wet suction method [2]. The percentage of moderate-to-high cellularity tissues was higher with conventional aspiration than with the wet suction method (conventional vs. wet, 77 vs. 61.5%). However, FNA needles were used in this study.

EUS-FNA of colorectal SELs can be performed, but inserting an echoendoscope into the deep colon is difficult. However, EUS-FNA has been successfully used for deep colorectal lesions by inserting an OV echoendoscope with an overtube and a guidewire [22, 23] or a novel FV echoendoscope with a guidewire [87, 88].

EUS-FNA for SELs smaller than 20 mm

In the case of SELs, the stomach is the most important target organ for EUS-FNA due to the highest incidence of GISTs. Because most gastric SELs of < 20 mm are considered benign or low malignancy, and because collecting specimens via EUS-FNA is technically difficult, EUS-FNA has been recommended for gastric SELs of > 20 mm.

However, metastatic cases of gastric GISTs of < 20 mm have been reported [89,90,91], and the Japanese guidelines 4th edition for the treatment of GISTs in 2022 weakly recommend surgery for gastric GISTs of < 20 mm [92]. Therefore, considering that a hypoechoic SEL of gastric muscular origin without delle cannot be diagnosed, endoscopic tissue sampling is necessary for the diagnosis; even if the SEL is < 20 mm, surgical resection is performed if EUS-FNA results show a malignant tumor such as a GIST. Conversely, if the SEL is diagnosed as a benign tumor such as a leiomyoma or schwannoma, the patient can be followed up without unnecessary treatment. Surgical resection of GISTs does not require lymph node dissection, and diagnosis of GISTs at a small size allows for a reduction of the lesion size.

The accuracy of EUS-FNA for SELs of < 20 mm varied from 45 to 80%, including reports of FNB needles (Table 2) [25, 26, 33, 36, 37, 41, 42, 55, 61,62,63,64, 69, 86]. Among these articles, Akahoshi et al. [26] reported the largest number of cases in their study of 90 cases; however, the specimen collection rate was 62.2%, which was insufficient. This article was published in 2014, and only FNA needles were used. They noted that even in cases of inadequate specimen collection at the first EUS-FNA, the total specimen collection rate improved to 73.3% by repeating the EUS-FNA. In a retrospective study by Inoue et al., the accuracy after propensity score matching was significantly lower for SELs of ≤ 20 mm (≤ 20 vs. > 20 mm, 66.7% vs. 96.7%) [61]. In our study, the specimen collection rates for 140 gastric SELs that underwent EUS-FNA were 80% (28/35), 94.9% (94/99), and 100% (6/6) for specimens of < 20 mm, 20–49 mm, and ≥ 50 mm, respectively, i.e., it was lower for specimens of < 20 mm. Specimens were collected in 80% of patients with an SEL of < 20 mm and 85.7% (24/28) of patients with an SEL of 15–19 mm [93].

Table 2 EUS-FNA for SELs < 20 mm or ≤ 20 mm
Full size table

In a comparison of FNA and FNB needles for SELs of < 20 mm, Sekine et al. found no difference in accuracy at > 20 mm (FNA vs. FNB, 75% vs. 77.8%), but at < 20 mm, the accuracy of FNB needles was significantly higher (FNA vs. FNB, 72.7% vs. 100%) [55]. Nagai et al. also found that the specimen collection rate was significantly higher with the FNB needle when used as the historical control in SELs of ≤ 20 mm (FNA vs. FNB, 45.4% vs. 81.1%) [69]. The specimen collection rate was also higher with the FNB needle for SELs of ≤ 15 mm (FNA vs. FNB, 38.5% vs. 94.1%).

In summary, the specimen collection rate and accuracy of EUS-FNA for SELs of < 20 mm are insufficient; however, once a diagnosis is obtained, EUS-FNA is useful for determining the treatment strategy (Fig. 4). The specimen collection rate and accuracy of EUS-FNA for SELs of < 20 mm are expected to improve with use of new FNB needles, FV echoendoscopes, and other techniques.

Fig. 4
figure 4

EUS-FNA of a small gastric GIST. a small SEL was detected in the gastric fornix (white arrow). b EUS using an ultrasound probe revealed an 11-mm-sized hypoechoic lesion in the muscularis propria. c A forward-viewing echoendoscope with a transparent hood made of polyvinyl chloride was inserted at the tip of the scope. d The puncture was performed using a 22-G fork-tip needle. e Hematoxylin/eosin staining showed clusters of spindle-shaped cells. f C-kit staining was positive for tumor cells, and a diagnosis of GIST was made. EUS-FNA endoscopic ultrasound-guided fine-needle aspiration, GIST gastrointestinal stromal tumor

Full size image

Sample handling during on-site evaluation

Rapid on-site evaluation (ROSE) is a method to determine whether an evaluable specimen has been collected and to decide whether to terminate the puncture or perform an additional puncture during EUS-FNA (Fig. 5) [94, 95]. Several meta-analyses have shown that EUS-FNA with ROSE helps to improve the specimen collection rate and accuracy for pancreatic lesions [96,97,98], and the usefulness of EUS-FNA in SELs has also been reported [1, 99]. However, in real clinical practice, cytologists and cytology technologists in some facilities experience difficulties performing EUS-FNA, and endoscopists use these methods for performing ROSE [95, 100, 101]. Further, several methods of specimen confirmation performed by endoscopists, including macroscopic on-site evaluation (MOSE) [102,103,104], stereomicroscopic on-site evaluation (SOSE) [105,106,107,108,109], and visual on-site evaluation (VOSE) [110], have been reported. MOSE is a method to evaluate the macroscopically visible core (MVC), a white, elongated portion of a specimen extruded on a glass slide. VOSE is a method used to confirm MVC by directly placing the specimen in a formalin solution, whereas SOSE is a method used to perform MOSE under a stereomicroscope.

Fig. 5
figure 5

Specimen handling in EUS-FNA of SELs. a The specimen collected using an FNA or FNB needle was extruded onto a glass slide using a stylet. b The specimens were divided into two groups: one for cytology and the other for histology. One specimen for cytology was fixed dry and ROSE stained with Cyto-quick stain, and the other specimen was fixed wet and Papanicolaou stained. c The specimen for histological examination was placed in a formalin bottle. EUS-FNA endoscopic ultrasound-guided fine-needle aspiration, SEL subepithelial lesion, FNA fine-needle aspiration, FNB fine-needle biopsy; ROSE, rapid on-site evaluation

Full size image

Conversely, the significance of ROSE may differ in EUS-FNA with the FNB needle because obtaining a proper specimen is easier than with the FNA needle. The diagnostic performance of the FNB needle is reportedly higher than that of the FNA needle used with ROSE, even without ROSE [111]. In a meta-analysis of SEL by Facciorusso et al., no difference in the specimen collection rate was observed between FNA and FNB needles in EUS-FNA with ROSE (odds ratio, 1.60); however, the FNB needle was superior to the FNA needle in terms of the specimen collection rate without ROSE (odds ratio, 9.85) [70]. Therefore, the FNB needle provides sufficient diagnostic performance for SEL without ROSE. Regarding the number of punctures for SELs, Suzuki et al. [112] reported that 2–3 punctures are recommended for EUS-FNA using the FNB needle without ROSE, depending on the lesion location and needle type.

In the case of EUS-FNA with the FNA needle, ROSE contributes to the improvement of diagnostic performance; whereas with the FNB needle, ROSE has a slight additional effect on diagnostic performance because of excellent original specimen collection. However, Han et al. reported that the FNB needle for SELs with ROSE provided adequate specimens in a single puncture [57]. Therefore, the significance of ROSE in EUS-FNA of SEL cannot be denied, and the combined use of the FNB needle and ROSE can effectively reduce the number of punctures and yield a sufficient specimen. In particular, ROSE for SELs can be used to successfully confirm the presence or absence of spindle-shaped cells, suggesting that mesenchymal tumors should be immunostained.

The evidence for MOSE, SOSE, and VOSE as on-site evaluations other than ROSE is insufficient. However, a meta-analysis by Mohan et al. in 2022 (including pancreatic masses and lymph nodes in addition to SELs) revealed that EUS-FNA with MOSE combined with a new FNB, Franseen, and fork-tip needle showed a sufficient specimen collection rate [104]. The specimen collection rate and accuracy were 94.7% and 90.6%, respectively.

Other specimen handling techniques include the cell block method reported for EUS-FNA in SELs [86, 113].

Adverse events

Although performing EUS-FNA for SELs is considered safe, potential adverse events associated with the procedure include bleeding, perforation, and infection. In a meta-analysis of upper GI SELs, three serious adverse events (0.3%) occurred after performing 978 EUS-FNAs, including FNB and TCB needles [44]. All these adverse events occurred in patients in whom 19-G needles were used: two patients had sepsis [48] and one died due to multiple organ failure after complications caused by FNA of a large centrally necrotic GIST [43]. Regarding GIST, Takasumi et al. reported a case of intratumoral infection in a patient with a GIST who underwent EUS-FNA with a 22-G fork-tip needle for gastric SELs, which conservatively improved with oral antimicrobial administration (Fig. 6) [1]. As for infections, EUS-FNA of cysts is problematic because of post-puncture infection within the cyst, and care should be taken when performing EUS-FNA of bronchogenic cysts, which may be difficult to diagnose on EUS imaging [4]. Therefore, EUS-FNA of cysts in the GI wall should not be performed without drainage of cyst contents. In addition, EUS-FNA of colonic SELs should be performed after colonic lavage even if the cyst is not a cyst. Levy et al. reported that adverse events occurred in 20.5% of 502 patients who underwent EUS-FNA through the colon wall, including lesions other than SELs, with pain, bleeding, and infection as the most common events [114].

Fig. 6
figure 6

Intratumoral infection after EUS-FNA. a Contrast-enhanced CT before EUS-FNA shows an SEL of the extragastric wall development type (yellow arrow). b EUS-FNA was performed using a 22-G fork-tip needle using a forward-viewing echoendoscope, and a diagnosis of GIST was made. c The patient had abdominal pain and fever after EUS-FNA, and CT showed hypoabsorption inside the tumor and increased fatty tissue density surrounding the tumor (yellow arrow). A diagnosis of intratumor infection or abscess formation was made. EUS-FNA endoscopic ultrasound-guided fine-needle aspiration, CT computed tomography, SEL subepithelial lesion, GIST gastrointestinal stromal tumor

Full size image

Regarding bleeding after EUS-FNA, among 1,135 EUS-FNA cases of SELs based on the Japanese Diagnosis Procedure Combination database, five (0.4%) of them had severe bleeding requiring blood transfusion or endoscopic treatment [115]. Among the 908 EUS-FNA cases, including 56 SEL cases, 114 were on antithrombotic drugs, and Polmanee et al. reported four cases (3.5%) of post-EUS-FNA bleeding in patients on antithrombotic drugs [116]. Of the four cases, two continued to take antithrombotic agents, one had heparin replacement, and one was off antithrombotic agents when EUS-FNA was performed. However, none of the patients had serious bleeding. Inoue et al. reported that in 742 patients who underwent EUS-FNA, including 129 SELs, 131 (17.7%) were on antithrombotic drugs; however, only one patient with SEL-like gastric cancer had intraoperative bleeding and no postoperative bleeding [117].

Furthermore, there are no reports of tumor seeding after EUS-FNA for SELs. Therefore, EUS-FNA for SELs can be performed safely; however, large blood vessels in the puncture route and puncturing through the GI tract wall should be avoided.

Conclusion

EUS-FNA has become the first choice for obtaining pathological tissue samples from GI SELs. With the advent of FNB needles, its diagnostic accuracy for GI SELs is improving. Furthermore, an FV echoendoscope and ROSE have also helped improve diagnostic accuracy. Therefore, EUS-FNA is currently performed aggressively even for SELs of < 20 mm. However, the difference between using EUS-FNA and MIAB should be discussed in the future.