Resection of the gastric submucosal tumor (G-SMT) originating from the muscularis propria layer: comparison of efficacy, patients’ tolerability, and clinical outcomes between endoscopic full-thickness resection and surgical resection

This study was approved by the institutional review board (First Affiliated Hospital, School of Medicine, Zhejiang University). The clinical records of a consecutive series of 90 patients, who underwent EFTR for primary G-SMTs at the department of gastroenterology between September 2011 and May 2019, were collected. 62 patients were enrolled into the EFTR group (study group) after exclusion based on the following criteria: (1) age < 18 or > 80; (2) the largest diameter of the target tumor < 1.0 cm; (3) gastric cancer or other GI diseases requiring treatment; (4) history of GI surgery or altered GI anatomic structures; (5) target gastric lesions originating from the submucosal layer; (6) multiple gastric tumors; (7) serious comorbidities; (8) patients undergoing LECS; and (9) the absence of EUS data (Fig. 1). For comparative assessment, a group of 62 patients with G-SMTs treated by surgery in the same period was matched for patient baseline characteristics, tumor clinicopathological features as well as the year part of the procedure date (Fig. 1). Overall, there were 62 and 62 patients in the EFTR and surgery group, respectively. To comply with the research ethics and the Personal information Protection Act, the included patients were replaced with surrogate numbers when we analyzed the data.

Prior to the EFTR or surgery, all patients were examined by ultrasonography and/or computed tomography (CT) and/or magnetic resonance imaging (MRI) of the abdomen to exclude tumor metastasis. Endoscopic ultrasonography (EUS) was also performed to determine the tumor size, originating layer, internal echogenicity, and growth pattern of target G-SMTs. The patients were followed up until September 15th, 2019 or death or tumor recurrence/metastasis requiring treatment or lost to follow-up, depending on which came first.

EFTR, as a novel and technically demanding procedure, allows for excision of a small piece of the complete (full-thickness) GI wall by using per-oral endoscopy. This resection maneuver is considered suitable for G-SMTs smaller than 30 mm and without ulceration, while the large G-SMTs (≥ 30 mm) were generally indicated for surgical resection [12, 17, 18]. However, the present study demonstrated that some large tumors (with ulcerations) could be removed by EFTR based on thorough preprocedural examinations with the assistance of abdominal CT, MRI, and EUS. The determination of the resection method for G-SMT was based on the overall consideration of the tumor characteristics, the operators’ experience, and the patients’ preferences. Maintaining the intact tumor capsule and minimizing the injuries to the surrounding normal tissue were prioritized during the resection.

All procedures were performed by the same team of experienced endoscopists (F.J, L.H.C., H.T.C. C.H.Y.) who had completed more than 1500 endoscopic treatment sessions within the upper GI tract, including ESD, endoscopic mucosal resection (EMR), per-oral endoscopic myotomy (POEM), Submucosal tunneling endoscopic resection (STER). The collaboration of two trained nurses or technicians and one anesthetist was also required.

After the patient was kept nil per os (NPO) for at least 6 h, propofol (1.5–2.0 mg/kg) was administered, whereas the general anesthetic with endotracheal intubation was selected for the anxious/agitated patients. During the whole procedure, the carbon dioxide (CO₂) insufflation of the peritoneal cavity and a double-channel upper gastrointestinal endoscopy (GIF-2TQ260M, Olympus, Tokyo, Japan) were applied. The basic operative steps are as follows: (1) marking dots were pointed circumferentially 2 mm away from margin of the target lesion with an electrosurgical knife; (2) a mixture solution (0.9% normal saline/10% glycerin fructose plus epinephrine (1:10,000) plus indigo carmine) was injected into the submucosal layer to facilitate submucosal elevation; (3) with appropriate choice of knives including Dual knife (KD-650L/U/Q, Olympus), Hook knife (KD-620LR, Olympus), and hybrid knife (20150-060/-300, Erbe Elektromedizin, Tübingen, Germany), a circumferential pre-cutting around the lesion was carried out deeply enough to gain access to the submucosal space. In some cases, the mucosa and submucosa covering the lesion are dissected in order to assess the tumor margin and to improve the operation view; (4) the MP tissue associated with the lesion was separated by hybrid knife or the alternate cutting with insulated-tip (IT) knife (KD-611L, Olympus) and Hook/Dual knife; (5) after the lesion was almost exposed and the remnant attached tissue was less than one-fifth, the lesion with the surrounding gastric tissue was resected by snaring method (NOE 342216-G, ENDO-FLEX, Voerde, Germany; M00562670, Boston Scientific Corporation, Natik, USA), and an active perforation would be created; (6) the gastric wall defect was approximated and closed by metallic clips (M00522610, Boston Scientific Corporation), or purse-string suturing method (loop-and-clip closure technique) using metallic clips and endoloops (LeCamp™ Loop-20/-30, LeoMed, Changzhou, China). For larger defects, the over-the-scope-clip (OTSC, 100.12, Ovesco Endoscopy, Tübingen, Germany) could also be opted (Fig. 2) [10]; (7) the lesion was pulled towards the gastric cavity and retrieved through the mouth; (8) a 20-gauge needle was inserted into the right upper quadrant to relieve the pneumoperitoneum; and (9) a nasogastric (NG) tube was routinely placed for GI decompression.

For G-SMT resection, either the laparoscopic wedge resection or the extended resection (e.g., total gastrostomy) is preferred depending on the size and anatomic site of the G-SMT [19]. The target gastric lesion at the gastric body or fundus was managed with wedge or sleeve-type resection. The localized lesion along with the associated gastric tissue was dissected using the linear gastrointestinal stapler (s) (EC60, Johnson & Johnson, New Jersey, USA) or the electrocautery device with a surgical margin of 10–20 mm, and then, it was put into a specimen bag and exteriorized through the trocar. The gastric wall defects were closed with a linear stapler(s) or hand-sewn. For larger lesions (≥ 50 mm) or lesions at specific locations (proximal to or at the gastroesophageal junction or cardia or antrum) [18, 20], the subtotal or distal gastrectomy with gastrojejunal or gastroduodenal reconstruction would be recommended [20].

A thorough screening for metastatic signs in the abdomen was also required during the surgery. Any suspicious lesions on the peritoneum would be resected (if possible) or biopsied for further staging of the tumor. The resection of any enlarged lymph nodes along with the primary lesion was also pursued. For the tumors with extension or invading into the adjacent organs, a more invasive resection should be performed to ensure tumor en bloc resection.

The specimen was cut open along the suture lines and measured with a millimeter ruler. Then it was immersed in 10% formalin solution. The sectioned tumor slices were stained with hematoxylin and eosin (H&E), and the number of mitotic figures per 50 high-powered fields was counted. The gastric GIST was confirmed by the immunohistochemical analysis of CD-117 (c-KIT), CD-34, DOG-1, SMA, Desmin, S-100 along with genotyping of KIT or PDGFRA. The risk classification of GISTs referred to the revised NIH grading system by Joensuu [21].

Patients were kept NPO for at least 48 h until occurrence of anal exsufflation or withdrawal of the NG tube. Then the patients were on liquid diets for the following 2–3 days and gradually returned to a normal diet over 1 week. Antibiotics (piperacillin–tazobactam 4.5 g or ceftriaxone 2 g three times daily plus ornidazole 500 mg twice daily), proton pump inhibitors (PPI, esomeprazole or pantoprazole 40 mg twice daily), hemostatics were administered intravenously. Then the PPI medications were orally taken for another 4 weeks after patient discharge.

(1) Tumor size The size was determined as the length in the largest dimension of the resected tumor. (2) Procedural success The target G-SMT was completely resected (en bloc or in piecemeal) by EFTR or surgery (laparoscopy or open surgery), and the gastric wall defect was successfully closed during the same procedure. (3) Histological confirmed R0 resection The dissected specimen without tumor residue at the resection margin was confirmed based on the histological examination. (4) Blood loss During the whole resection process, the amount of bleeding was estimated by the volume of blood collected in the suction tank and/or weighing the gauze swabs collecting the blood.

The follow-up strategy depended on pathological reports of the G-SMT. Approximately 1 month after G-SMT resection by EFTR, the GI endoscopy was conducted to observe postprocedural healing and exclude residual tumors. The surveillance endoscopy was then recommended for the patients every 3–6 months, and the re-examination interval could be prolonged to 1 year after 3 endoscopic sessions. The patients who underwent surgery received first endoscopy at 3 months postoperatively and then repeated semiannually. All patients were periodically (every 6 months–1 year) followed up through an outpatient visit or telephone call to monitor their oncological outcomes. The re-examination records of endoscopy, EUS, CT, or MRI were collected and reviewed.

The SPSS 19.0 statistical software (SPSS, Inc., Chicago, IL, USA) was used to set up the database and analyze the results. Measured data were expressed as the mean value ± standard deviation (SD). Differences between two mean values were estimated by an unpaired Student t test. The categorical variables were presented as frequencies and corresponding percentages, and the Pearson Chi-square or Fisher’s exact test was used for comparison. The parameters relative to number of days (follow-up duration, hospital stay, postoperative bowel function restoration, fasting, antibiotics usage) were evaluated using the Wilcoxon Mann–Whitney test and quantified by the median with range. All statistical tests were two-sided, and p < 0.05 was considered significant.