Endoscopic vs laparoscopic resection for gastric gastrointestinal stromal tumors: Oncological outcomes

Abstract

BACKGROUND

Surgical resection is the core treatment for localized gastric gastrointestinal stromal tumors (LGISTs). Advances in minimally invasive techniques have led to the use of both endoscopic and laparoscopic resections; however, there is controversy regarding their oncological efficacy and safety, especially due to the lack of head-to-head comparative data with balanced baselines.

AIM

To systematically compare the perioperative outcomes and mid-term oncological efficacy of endoscopic vs laparoscopic resection for LGISTs, and provide an evidence-based reference for clinical surgical approach selection.

METHODS

Patients with LGIST who underwent surgery in our hospital between January 2023 and January 2024 were retrospectively enrolled. After 1:1 propensity score matching, 45 patients who received endoscopic resection were assigned to the endoscopic group, and 45 patients who underwent laparoscopic resection were included in the laparoscopic group. Intraoperative indicators (such as operation time, blood loss, R0 resection rate), postoperative recovery indicators (including hospital stay, time to first flatus), complication rate, and mid-term oncological outcomes [1-year/3-year recurrence-free survival (RFS), overall survival (OS)] were compared between the two groups. Multivariate Cox regression was used to identify prognostic factors.

RESULTS

After matching, the baseline data of the two groups were comparable (P > 0.05). The endoscopic group was superior to the laparoscopic group in terms of operation time [80 minutes (65-100 minutes) vs 95 minutes (80-120 minutes), P = 0.002], intraoperative blood loss [25 mL (15-35 mL) vs 55 mL (40-90 mL), P < 0.001], and postoperative hospital stay [5 days (4-7 days) vs 7 days (6-9 days), P < 0.001]. There were no significant differences in the rates of R0 resection (95.6% vs 97.8%, P = 0.617), intraoperative tumor rupture (2.2% vs 4.4%, P = 1.000), and 30-day postoperative complications (11.1% vs 22.2%, P = 0.152) between the two groups. With a median follow-up of 32 months, the 3-year RFS (93.3% vs 91.1%, P = 0.695) and 3-year OS (97.8% vs 95.6%, P = 1.000) rates were comparable between the two groups. Multivariate analysis showed that tumor size (HR = 1.38, P = 0.002), mitotic count (HR = 1.18, P = 0.010), and National Institutes of Health risk stratification (intermediate risk vs low risk: HR = 5.12, P = 0.001) were independent risk factors for RFS, while surgical approach was not an independent prognostic factor (P = 0.558).

CONCLUSION

In carefully selected LGIST cases, endoscopic resection achieves comparable mid-term oncological efficacy to that of laparoscopic resection, while offering the advantages of shorter operation time, less blood loss, and faster postoperative recovery. It can therefore be a minimally invasive treatment option for eligible patients, with surgical decision-making based on tumor characteristics and multidisciplinary assessments.

Key Words: Endoscopy; Laparoscopy; Localized gastric gastrointestinal stromal tumor; Complications; Prognosis

Core Tip: This study compared the efficacy of endoscopic and laparoscopic resection for localized gastric gastrointestinal stromal tumors (LGISTs) using propensity score matching to control for baseline bias. Endoscopic resection was found to be superior in terms of reduced operation time, blood loss, and postoperative hospital stay, while achieving comparable mid-term oncological outcomes (3-year recurrence-free survival and overall survival) to laparoscopic resection. Tumor size, mitotic count, and National Institutes of Health risk stratification were key factors affecting prognosis, and the surgical approach had no direct influence on recurrence risk. The study provides evidence-based support for “prioritizing minimally invasive endoscopic treatment for carefully selected LGIST patients”.

INTRODUCTION

Gastrointestinal stromal tumor (GIST), the most common mesenchymal tumor of the gastrointestinal tract, occurs most frequently in the stomach (50%-63%) and small intestine (29%-35%)[1-3]. Most GISTs are sporadic and, apart from several rare tumor syndromes, no risk factors have been identified[4]. The clinical manifestations of GIST vary. Some patients are asymptomatic, leading to incidental detection of the tumor during abdominal imaging examinations, surgeries for unrelated conditions, or physical palpation[5], while others may present with clinical events such as gastrointestinal bleeding or perforation. The diagnosis of GIST relies on initial endoscopic examination, followed by histopathological and immunohistochemical confirmation in tissue samples obtained during biopsy or tumor resection[6,7]. Currently, all GISTs are assumed to be potentially malignant, although most exhibit an indolent clinical course[8]. The risk of GIST recurrence after resection is most commonly predicted using the modified National Institutes of Health (NIH) scoring system[9], in which recurrence risk is assessed based on tumor size, mitotic count, and location. Specifically, the system classifies gastric GISTs into very low-risk [< 2 cm with < 5 mitoses per 50 high-power fields (HPFs)], low-risk (2-5 cm with < 5/50 HPFs), intermediate-risk (< 5 cm with 6-10/50 HPFs or 5-10 cm with < 5/50 HPFs), and high-risk (> 5 cm with > 5/50 HPFs or > 10 cm with any mitotic count) tumors based on tumor size and mitotic count.

Localized GIST (LGIST) is the most common form of GIST seen on initial presentation. These tumors are confined to the wall of the primary gastrointestinal tract with possible restricted invasion of adjacent tissues (e.g., perigastric fat, small intestinal mesentery), with no distant metastasis (such as liver or lung metastasis), peritoneal seeding, or major vascular invasion. This GIST subtype responds better to treatment and is associated with a more favorable prognosis. While surgical resection remains the mainstay of LGIST treatment, the optimal surgical approach is disputed and is currently a focus of clinical research. The development of minimally invasive techniques has expanded the surgical options from traditional open surgery to laparoscopic and endoscopic resection. Surgical resection for LGISTs with tumor diameters > 2 cm is recommended in guidelines worldwide[10,11]. Among these approaches, laparoscopic surgery represents a standard minimally invasive procedure for tumors < 5 cm in diameter located at “favorable sites”, such as the anterior gastric wall and greater curvature. In addition, endoscopic techniques, including endoscopic submucosal dissection (ESD) and endoscopic full-thickness resection (EFTR), have demonstrated significant potential in treating small GISTs (especially those ≤ 2 cm) due to their unique advantages of using natural orifices for insertion, thereby avoiding trauma to the abdominal wall. These advantages have led to increased acceptance and application of these techniques[12]. However, the use of endoscopic techniques in the treatment of large-diameter LGISTs (e.g., 2-5 cm) remains highly controversial. The core of the controversy lies in whether endoscopic resection can achieve oncological radicality (R0 resection rate) and long-term recurrence-free survival (RFS) rates equivalent to those of laparoscopic surgery for larger and more complex tumors, as well as their ability to control the risks of intraoperative perforation and tumor rupture[13]. Moreover, most existing studies have involved small samples and may also suffer from significant selection bias (e.g., the typically smaller volumes of tumors in patient groups treated using endoscopy), resulting in a lack of high-level evidence. Therefore, in the era of minimally invasive surgery, the choice between laparoscopic and endoscopic techniques based on tumor size, location, and growth pattern, in terms of providing the most appropriate, effective, and safe individualized surgical treatment for patients with LGIST, represents an urgent problem. This study aims to systematically compare the perioperative outcomes and mid-term efficacy of endoscopic and laparoscopic resection of LGIST using a rigorous propensity score matching (PSM) analysis, to provide an evidence-based reference for optimizing clinical decision-making.

MATERIALS AND METHODS

Study design

This single-center, retrospective study consecutively enrolled patients with LGIST who had undergone surgical treatment at JiuJiang No. 1 People’s Hospital, Jiujiang City Key Laboratory of Cell Therapy Hospital between January 1, 2023, and January 1, 2024. Patients were divided into the endoscopic resection (ESD/EFTR) and laparoscopic surgery groups based on the surgical procedure performed.

Study population

Inclusion criteria: (1) Definitive diagnosis of primary gastric GIST confirmed by postoperative pathological analysis of paraffin sections or preoperative endoscopic ultrasound (EUS)-guided fine-needle aspiration biopsy (confirmed by postoperative pathology), with positive immunohistochemical staining for CD117 and/or DOG1; (2) Preoperative imaging and EUS assessments consistent with American Joint Committee on Cancer T1-T2 staging, tumor diameter of 1.0-10.0 cm, distance from the cardia or pylorus of ≥ 3.0 cm, and originating from the submucosa or superficial muscularis propria of the gastric wall; (3) Pathological risk stratification of very low, low, or intermediate risk according to the National Comprehensive Cancer Network (NCCN) guidelines, based on tumor size and mitotic count (in 50 HPFs); (4) Underwent pure endoscopic resection (ESD/endoscopic submucosal excavation) or local laparoscopic resection, with no combination of the two procedures; (5) Preoperative American Society of Anesthesiologists (ASA) physical status classification of I-II, with no severe cardiopulmonary, hepatic, or renal dysfunction, coagulation disorders, or other surgical contraindications; and (6) Complete records of postoperative 30-day complications and 1-year/3-year recurrence during follow-up available in the medical records system or follow-up registry.

Exclusion criteria: (1) Presence of distant metastasis, invasion of adjacent organs, or serosal penetration; tumor diameter < 1.0 cm/> 10.0 cm, or distance from the cardia/pylorus < 3.0 cm; (2) Previous treatment with targeted agents (e.g., imatinib), argon plasma coagulation (APC), radiofrequency ablation, or other therapies before surgery; (3) History of abdominal surgery, or the presence of multiple gastric GISTs; (4) Comorbidity with other types of malignant tumor (including past history), ASA class III-IV diseases, acute infections, or psychiatric disorders potentially influencing recovery or follow-up; (5) Absence of key data in the medical records, follow-up duration < 1 year, or incomplete information on recurrence/complications; and (6) Age < 18 years, or inability to cooperate with follow-up due to cognitive/Language impairment or medical disputes.

Ultimately, a total of 103 patients meeting the inclusion and exclusion criteria were enrolled in the analysis.

Treatment methods

All patients were evaluated before surgery by a multidisciplinary team (MDT) consisting of gastrointestinal surgeons, gastroenterologists, oncologists, and radiologists. The surgical plan was determined collectively based on the tumor size, location (distance from the cardia/pylorus), growth pattern (intraluminal/exophytic), and estimated surgical risk.

Preoperative preparation: Routine preoperative preparation was undertaken for both groups: (1) Fasting for 12 hours and nil by mouth for 4 hours before surgery; (2) Patients undergoing EFTR or laparoscopic surgery received bowel preparation by oral administration of polyethylene glycol electrolyte powder one day before surgery; (3) Antiplatelet and anticoagulant drugs were discontinued seven days before surgery, while low-molecular-weight heparin bridging therapy and low-molecular-weight heparin were stopped 12 hours before surgery; (4) Prophylactic second-generation cephalosporins (e.g., cefuroxime) were administered by intravenous infusion 30 minutes before surgery for patients undergoing EFTR or laparoscopic surgery; and (5) A nasogastric tube was positioned before surgery in the laparoscopic group, but not in the endoscopic group.

Surgical procedures in the endoscopic group: The surgical approach was selected based on EUS findings.

The use of ESD was indicated for tumors that showed predominantly intraluminal growth with no tight adhesion to the serosa. The key steps were as follows: (1) A mixed solution (glycerin fructose-indigo carmine-epinephrine) was administered submucosally to create a submucosal cushion with a lifting sign height ≥ 5 mm; (2) A circumferential mucosal incision was made 5 mm from the edge of the lesion after marking; (3) The submucosal layer was dissected for complete resection of the tumor, avoiding damage to the muscularis propria; and (4) Hemostasis of visible vessel stumps on the wound surface was performed using hot biopsy forceps or APC.

The use of EFTR was indicated for tumors with transmural growth or tight adhesion to the serosa. The key steps were as follows: (1) The submucosal injection and marking steps were performed as described for ESD above; (2) After circumferential mucosal incision, dissection was continued until the creation of an intentionally controlled perforation (diameter ≤ 2 cm) of the serosa, enabling complete full-thickness resection of the tumor; and (3) The wound was closed with endoclips and nylon loops using the “purse-string suture” technique to ensure no air leakage.

The surgical team involved in this study had extensive experience in minimally invasive endoscopic surgery. Each surgeon performed more than 150 ESD/EFTR procedures annually, including over 50 cases of gastric GIST resection. The center conducts approximately 120-150 endoscopic resections for gastric GIST each year, with a technical success rate of over 95% for complex cases (such as tumors in unfavorable locations or with transmural growth).

Surgical procedures in the laparoscopic group: Patients were placed in a supine position, and general anesthesia was administered, followed by establishment of a CO₂ pneumoperitoneum (intra-abdominal pressure maintained at 12 mmHg). A standard five-port technique was used (a 10 mm umbilical observation port, and four 5 mm working ports at other sites). After exploration of the abdomen to confirm the location of the tumor and its relationship with surrounding blood vessels, the surgical approach was selected based on the tumor size and location: (1) For tumors < 5 cm in diameter located in the gastric body or fundus, local laparoscopic resection was performed, in which the gastric wall was incised 1 cm from the tumor edge using an ultrasonic scalpel, after which the tumor was completely resected before layered suturing of the gastric wall; and (2) For tumors ≥ 5 cm in diameter or located in the gastric antrum, laparoscopic wedge resection was performed. This involved the use of a linear cutting stapler to transect the gastric wall and tumor 1 cm from the tumor edge, ensuring a margin of ≥ 1 cm from the tumor edge. All resected tissue was placed in a retrieval bag and extracted through an enlarged trocar site. R0 resection was the primary goal for all surgeries.

Postoperative management: All patients received perioperative care based on the enhanced recovery after surgery (ERAS) protocol. This involved: (1) Early postoperative mobilization (bedside activity at 6 h postoperatively, ambulation at 24 hours postoperatively); (2) Initiation of a liquid diet 24 hours (endoscopic group) or 48 hours (laparoscopic group) after surgery, with a gradual transition to a semi-liquid and soft diet; and (3) Pain assessment using the visual analog scale (VAS); patients with a VAS score ≥ 4 received intravenous flurbiprofen axetil for analgesia.

Postoperative evaluation and adjuvant therapy: (1) Assessment of complications: Complications [bleeding (hematemesis/melena or hemoglobin decrease ≥ 20 g/L), perforation (abdominal pain with free intraperitoneal air), and infection (body temperature ≥ 38.5 °C and white blood cell count ≥ 12 × 10⁹/L)] were recorded and graded according to the Clavien-Dindo classification system, along with the management strategies applied; (2) Pathological assessment: Two senior pathologists independently evaluated tumor size, mitotic count (/50 HPFs), and resection margin status, stratifying risk according to the NIH criteria; and (3) Molecular testing: Sanger sequencing of c-KIT (exons 9, 11, 13, 17) and PDGFRA (exons 12, 18) was performed on specimens from patients classified as being of intermediate and high risk as a guide for adjuvant therapy.

Assessments by the MDT were conducted based on the results of the postoperative NIH risk stratification and molecular testing. All high-risk patients received adjuvant therapy with imatinib mesylate. Treatment dosage and duration were adjusted according to the type of gene mutation: A dosage of 400 mg/day for at least three years was recommended for patients with mutations in exon 11 of c-KIT, while an initial dosage of 600-800 mg/day was used for those with mutations in exon 9. Patients with the D842V mutation in PDGFRA, who are inherently resistant to imatinib, were not recommended for adjuvant imatinib therapy. All treatment recommendations were implemented after full communication with the patients and their families.

Outcome measures

Demographic and clinical data included information on sex, age, body mass index (BMI), initial symptoms, preoperative comorbidities, tumor size, mitotic count, tumor location, growth pattern, and cytological morphology. Tumor location was categorized into “favorable sites” (including the gastric fundus, anterior wall of the gastric body, and greater curvature) and “unfavorable sites” (including the esophagogastric junction, lesser curvature of the gastric body, posterior wall of the gastric body, gastric antrum, and pylorus) according to the NCCN guidelines and relevant literature. Tumor growth patterns were classified as “intraluminal” (main body of the tumor protruding into the gastric lumen) or “exophytic” (main body of the tumor protruding outside the gastric lumen) based on preoperative imaging and EUS assessments.

Intraoperative outcomes: These included the operation time, intraoperative blood loss (estimated by suction volume + gauze blood absorption in the endoscopic group, and by the difference between the irrigation fluid volume and suction fluid volume + gauze blood absorption in the laparoscopic group), R0 resection rate, rate of intraoperative tumor rupture, and rate of conversion to open surgery (laparoscopic group).

Postoperative recovery outcomes included the time to first flatus, time to first oral intake of liquid, time to ambulation, and length of postoperative hospital stay.

Postoperative complications: All complications occurring within 30 days after surgery were recorded, and their severity was graded using the Clavien-Dindo classification system. Focus was placed on the incidence of bleeding, perforation, intra-abdominal infection, and pulmonary infection.

Oncological outcomes represented the 1-year and 3-year rates of RFS and overall survival (OS).

Postoperative follow-up

All patients were followed up postoperatively through telephone calls. The follow-up time points were set at 1 months, 3 months, 6 months, 12 months, 18 months, and 24 months after surgery. The information obtained included the patient's survival status, symptoms, and occurrence of complications, all of which were included in the electronic medical record system. Abdominal enhanced computed tomography was performed at 6 months, 12 months, and 24 months postoperatively to assess tumor recurrence. The follow-up period ended on September 1, 2025.

Subgroup classification criteria

Tumor location: Based on the NCCN guidelines and published research, the gastric anatomical sites were divided into favorable and unfavorable site groups. Tumors included in the favorable site group were those located in the gastric fundus, anterior wall of the gastric body, and greater curvature, while the unfavorable site group comprised tumors located in the esophagogastric junction, lesser curvature of the gastric body, posterior wall of the gastric body, gastric antrum, and pylorus. In the laparoscopic group, 23 cases were included in the favorable site subgroup, with 22 cases in the unfavorable site subgroup, while in the endoscopic group, 23 cases were classified as being in favorable sites and 22 as being located in unfavorable sites.

Tumor growth pattern: Based on the relationship between the main body of the tumor and the gastric wall, tumors were divided into intraluminal and exophytic groups. The laparoscopic group included 23 cases of intraluminal tumors and 22 cases of exophytic tumors, while the endoscopic group contained 25 cases of intraluminal tumors and 20 of exophytic tumors.

Statistical analysis

Data were analyzed using SPSS 27.0 (IBM Corp., Armonk, NY, United States). Data distributions were analyzed, and data conforming to a normal distribution are presented as mean ± SD and were compared using independent-samples t-tests. Non-normally distributed continuous data are presented as median (interquartile range) and compared using the Mann-Whitney U test. Categorical variables are expressed as n (%), and intergroup comparisons were performed using the χ² test or Fisher's exact test. To control for selection bias, 1:1 PSM was performed for variables with inconsistencies in baseline characteristics between the two groups, using age, sex, BMI, tumor size, mitotic count, tumor location, growth pattern, and cytological morphology as covariates. Survival rates (OS and RFS) were analyzed using Kaplan-Meier curves, with intergroup comparisons using log-rank tests. The significance level was set at α = 0.05, and a P value < 0.05 was considered statistically significant.

RESULTS

Baseline characteristics of patients

A total of 123 eligible patients were initially screened. After 1:1 PSM, 90 patients were ultimately included in the analysis, with the endoscopic and laparoscopic groups each containing 45 patients. After matching, there were no statistically significant differences between the groups in terms of baseline data (including age, sex, BMI, tumor size, mitotic count, NIH risk stratification, tumor location, and growth pattern, all P > 0.05; Table 1). The significant selection bias present before matching was effectively controlled.

Table 1 Comparison of baseline characteristics between the two groups.

Variable	Pre-matching		P value	Post-matching		P value
	Endoscopic (n = 50)	Laparoscopic (n = 73)		Endoscopic (n = 45)	Laparoscopic (n = 45)
Demographic data
Age (years)	57.9 ± 10.1	59.8 ± 9.7	0.285	58.5 ± 9.8	58.9 ± 10.2	0.841
Gender (male)	28 (56.0)	40 (54.8)	0.892	25 (55.6)	24 (53.3)	0.829
BMI (kg/m2)	23.8 ± 2.9	24.3 ± 3.1	0.341	24.0 ± 2.8	24.1 ± 3.0	0.867
Tumor characteristics
Tumor size (cm)	2.8 (2.0-3.8)	4.3 (3.0-5.8)	< 0.001	3.5 (2.6-4.4)	3.6 (2.7-4.5)	0.723
Mitotic count (/50 HPF)	2 (1-4)	4 (2-6)	0.003	3 (1-5)	3 (2-5)	0.435
NIH risk stratification
Very low/low risk	35 (70.0)	38 (52.1)	0.009	28 (62.2)	26 (57.8)	0.654
Intermediate risk	15 (30.0)	35 (47.9)		17 (37.8)	19 (42.2)
Tumor location
Favorable sites	38 (76.0)	30 (41.1)	< 0.001	23 (51.1)	23 (51.1)	1
Unfavorable sites	12 (24.0)	43 (58.9)		22 (48.9)	22 (48.9)
Growth pattern
Intraluminal	35 (70.0)	35 (47.9)	0.002	25 (55.6)	23 (51.1)	0.832
Exophytic	15 (30.0)	38 (52.1)		20 (44.4)	22 (48.9)

Data were showed as n (%) or mean ± SD or median (interquartile range). BMI: Body mass index; NIH: National Institutes of Health; HPF: High-power field.

Notably, before matching, patients in the laparoscopic group had significantly larger tumor sizes (4.3 cm vs 2.8 cm, P < 0.001), higher mitotic counts (4/50 HPFs vs 2/50 HPFs, P = 0.003), higher proportions of intermediate-risk tumors (47.9% vs 30.0%, P = 0.009), and higher proportions of tumors located at unfavorable sites (58.9% vs 24.0%, P < 0.001) compared with the endoscopic group. This distribution was consistent with the clinical decision-making logic of surgical approach selection based on the complexity of the tumor.

Comparison of perioperative outcomes

Surgery was successfully completed in all patients, with no intraoperative deaths. As shown in Table 2, the endoscopic group exhibited significantly superior perioperative outcomes: Shorter operation time (80 minutes vs 95 minutes, P = 0.002), less intraoperative blood loss (25 mL vs 55 mL, P < 0.001), and shorter postoperative hospital stay (5 days vs 7 days, P < 0.001). The rates of R0 resection (97.8% vs 95.6%, P = 0.617) and intraoperative tumor rupture (2.2% vs 4.4%, P = 1.000) were comparable between the two groups. The overall complication rate was lower in the endoscopic group than that in the laparoscopic group (11.1% vs 22.2%), although the difference was not statistically significant (P = 0.152).

Table 2 Comparison of perioperative outcomes between the two groups.

Variable	Endoscopic (n = 45)	Laparoscopic (n = 45)	Test statistic	P value
Overall comparison
Operation time (minute)	80 (65-100)	95 (80-120)	Z = -3.112	0.002
Intraoperative blood loss (mL)	25 (15-35)	55 (40-90)	Z = -5.421	< 0.001
R0 resection rate	43 (95.6)	44 (97.8)	Fisher	0.617
Intraoperative tumor rupture	1 (2.2)	2 (4.4)	Fisher	1.000
Postoperative hospital stays (days)	5 (4-7)	7 (6-9)	Z = -4.225	< 0.001
Postoperative complications (Clavien-Dindo ≥ II)	5 (11.1)	10 (22.2)	χ2 = 2.051	0.152
Subgroup analysis by location
Favorable sites	n = 23	n = 23
Operation time (minute)	75 (60-90)	85 (70-100)	Z = -2.315	0.021
Intraoperative blood loss (mL)	20 (10-30)	50 (35-70)	Z = -4.521	< 0.001
Unfavorable sites	n = 22	n = 22
Operation time (minute)	90 (75-110)	110 (95-135)	Z = -2.987	0.003
Intraoperative blood loss (mL)	30 (20-45)	65 (50-105)	Z = -4.678	< 0.001
Subgroup analysis by growth pattern
Intraluminal	n = 25	n = 23
Operation time (minute)	70 (55-85)	90 (75-105)	Z = -3.224	0.001
Exophytic	n = 20	n = 22
Operation time (minute)	95 (80-115)	100 (85-125)	Z = -0.912	0.362

Data were showed as n (%) or median (interquartile range).

The results of the subgroup analysis further revealed the influence of the anatomical characteristics of the tumor on outcomes.

By tumor location: Irrespective of whether the tumor was located in a favorable or unfavorable site, the endoscopic group showed significant advantages in terms of operation time and intraoperative blood loss (both P < 0.05). Notably, surgery at unfavorable sites was associated with a significantly longer operative time and higher blood loss compared to surgery for tumors in favorable sites.

By growth pattern: For intraluminal tumors, the duration of surgery in the endoscopic group was significantly shorter than that in the laparoscopic group (70 minutes vs 90 minutes, P = 0.001), while for exophytic tumors, the operation time did not differ significantly between the two approaches (95 minutes vs 100 minutes, P = 0.362).

Postoperative complications

No significant difference in the overall complication rate (Clavien-Dindo ≥ grade II) was observed between the endoscopic and laparoscopic groups in the overall matched cohort (11.1% vs 22.2%, P = 0.152). Although a higher incidence of infectious complications (intra-abdominal infection and pulmonary infection) was seen in the laparoscopic group, the intergroup difference was not statistically significant. Among the three perforation cases, two were planned controlled perforations performed during EFTR (1 case in the endoscopic group and 0 cases in the laparoscopic group) that were successfully closed via endoscopic suturing, and one was an unexpected rupture that occurred during ESD (endoscopic group: 0 cases, laparoscopic group: 1 case) and was managed by conservative treatment. All complications were resolved after active intervention, and there were no severe complications of grade IV or above or deaths. Details are shown in Table 3.

Table 3 Comparison of postoperative complications between the two groups, n (%).

Complication type & management	Total (n = 90)	Endoscopic (n = 45)	Laparoscopic (n = 45)	P value
Overall complications (Clavien-Dindo ≥ II)	15 (16.7)	5 (11.1)	10 (22.2)	0.152
By grade
Grade II	10 (11.1)	4 (8.9)	6 (13.3)	0.506
Grade IIIa	4 (4.4)	1 (2.2)	3 (6.7)	0.617
Grade IIIb	1 (1.1)	0 (0.0)	1 (2.2)	1.000
Grade IV-V	0 (0.0)	0 (0.0)	0 (0.0)	-
By type
Bleeding - endoscopic hemostasis/transfusion	6 (6.6)	2 (4.4)	4 (8.8)	0.398
Perforation/fistula - conservative treatment/aspiration drainage	3 (3.3)	1 (2.2)	2 (4.4)	1.000
Intra-abdominal infection - antibiotics/aspiration drainage	3 (3.3)	2 (4.4)	1 (2.2)	1.000
Pulmonary infection - antibiotic treatment	4 (4.4)	1 (2.2)	3 (6.7)	0.617
Others	1 (1.1)	0 (0.0)	1 (2.2)	1.000

P values were derived from χ² tests or Fisher's exact tests. One patient may have more than one complication.

Survival analysis

All patients completed follow-up, with a median follow-up duration of 32 months. The 3-year RFS rates were 93.3% in the endoscopic group and 91.1% in the laparoscopic group (P = 0.695), while the 3-year OS rates were 97.8% and 95.6%, respectively (P = 1.000). Seven patients had experienced recurrence by the last follow-up (3 in the endoscopic group and 4 in the laparoscopic group), and 3 patients had died (see Figure 1 and Table 4).

Figure 1 Kaplan-Meier curves comparing long-term outcomes between endoscopic and laparoscopic resection for localized gastric gastrointestinal stromal tumor. A: Recurrence-free survival; B: Overall survival.

Table 4 Comparison of survival outcomes between the two groups.

Survival outcome	Endoscopic (n = 45)	Laparoscopic (n = 45)	Test statistic	P value
Median follow-up (months)	32 (26-38)	32 (26-38)	-	0.935
Recurrence events, n (%)	3 (6.7)	4 (8.9)	-	0.695
Death events, n (%)	1 (2.2)	2 (4.4)	-	1.000
RFS rate (%)
1-year	100	100	χ2 = 0.152	0.695
3-year	93.3	91.1
OS rate (%)
1-year	100	100	χ2 = 0.000	0.991
3-year	97.8	95.6

Survival rates were calculated using the Kaplan-Meier method. P values were derived from log-rank tests. RFS: Recurrence-free survival; OS: Overall survival.

Multivariate analysis of prognostic factors

To identify independent risk factors for recurrence, multivariate analysis was conducted by including multiple variables in a Cox proportional hazards model. The results showed that tumor size (HR = 1.38, 95%CI: 1.12-1.70, P = 0.002), mitotic count (HR = 1.18, 95%CI: 1.04-1.34, P = 0.010), and NIH risk stratification (intermediate-risk vs low-risk: HR = 5.12, 95%CI: 1.88-13.93, P = 0.001) were independent risk factors for RFS. In contrast, the surgical approach (laparoscopic vs endoscopic) was not an independent prognostic factor (P = 0.558). It should be noted that the conclusion of the surgical approach not being an independent prognostic factor is preliminary due to the small sample (90 cases) and low levels of recurrence (7 cases); larger cohorts and longer follow-up are needed to confirm this. Details are provided in Table 5.

Table 5 Multivariate Cox regression analysis of factors associated with recurrence-free survival.

Variable	HR	95%CI	P value
Surgical approach (laparoscopic vs endoscopic)	1.32	0.52-3.35	0.558
Age (per 1-year increase)	1.01	0.96-1.07	0.668
Tumor size (per 1-cm increase)	1.38	1.12-1.70	0.002
Mitotic count (per 1 increase/50 HPF)	1.18	1.04-1.34	0.010
NIH risk stratification (intermediate vs low)	5.12	1.88-13.93	0.001
Tumor location (unfavorable vs favorable)	1.65	0.60-4.52	0.328

The model was adjusted for surgical approach, age, tumor size, mitotic count, National Institutes of Health risk stratification, and tumor location. NIH: National Institutes of Health; HPF: High-power field.

DISCUSSION

GISTs differ fundamentally from gastric adenocarcinomas in terms of their biological behavior. Most LGISTs exhibit expansive growth with an extremely low rate of lymph node metastasis, with their prognosis depending primarily on the type of genetic mutation involved and their NIH risk stratification, rather than on the extent of lymphadenectomy. This characteristic indicates that the core goal of surgery is the achievement of complete resection with negative margins (R0 resection), rather than extensive lymphadenectomy[14]. Therefore, on the premise of ensuring negative margins, endoscopic resection, which is theoretically less invasive, can lead to oncological outcomes comparable to those of laparoscopic surgery. Moreover, endoscopic resection reduces surgical trauma and speeds postoperative recovery due to its advantage of transluminal surgery via natural orifices. There is currently no consensus on the optimal treatment strategy for LGISTs, with controversy surrounding “the selection and applicable scope of minimally invasive surgical approaches”. Existing guidelines generally recommend the use of surgical resection for LGISTs with diameters > 2 cm. Among the varied surgical options, open surgery is recognized as the first-line approach, while laparoscopic surgery, due to its relatively less invasive nature, is restricted to tumors < 5 cm in diameter located at “favorable sites” such as the gastric fundus or gastric body[14-16]. The development of endoscopic resection enables further reductions in surgical trauma. The relatively low difficulty associated with LGIST resection (due largely to their expansive growth and extremely low rates of lymph-node metastasis), as well as their low potential for malignancy (malignancy rate of 1.9%-16%)[17], may provide a biological foundation for the use of endoscopic resection. Endoscopic techniques can also, due to their application of transluminal operation through natural orifices, overcome the limitation of “intraoperative localization difficulty” encountered by laparoscopic surgery in treating tumors at unfavorable sites (such as the gastric antrum and cardia) or intraluminal lesions, and can thereby theoretically accelerate postoperative recovery while ensuring therapeutic efficacy.

However, endoscopic resection has not yet been fully accepted nor routinely applied in clinical practice. On one hand, there is widespread concern about its oncological safety, as, compared with laparoscopic or open surgery, it is more difficult to achieve complete (R0) resection using endoscopy. Incomplete resection can occur due to the limited operating space, especially when the tumor is associated with transmural growth or adhesion to the serosal layer, thereby increasing the risk of peritoneal seeding of tumor cells. On the other hand, technique-related challenges of endoscopic resection have also been described, including the relatively high incidence of intraoperative bleeding and perforation, as well as a lack of comparative data on long-term (over five years) tumor recurrence and survival, resulting in insufficient verification of its medium- and long-term safety[14-16]. This conflict between the theoretical advantages of its minimal invasiveness and practical clinical concerns poses a dilemma for clinical decision-making for LGISTs. Specifically, there is no consensus or unified standard on the necessity of surgery (e.g., whether immediate resection is required for some small tumors) nor on the selection of surgical approach (endoscopic/Laparoscopic/open surgery); these discrepancies between clinical practice and guideline consensus require urgent resolution. To address this issue, the present study retrospectively analyzed the clinical data of patients with LGIST who underwent either endoscopic or laparoscopic resection in our hospital, and used PSM to control for differences in baseline characteristics (such as tumor size, location, and risk stratification) between the two groups. This approach provided head-to-head comparative data between the two surgical approaches, with the intention to objectively evaluate the short-term recovery advantages and medium-term oncological efficacy of endoscopic resection, and thus offer an evidence-based basis for clarifying the controversy surrounding the treatment of LGISTs and optimizing clinical decision-making.

This study comprehensively compared the short- and long-term outcomes of endoscopic and laparoscopic resection for LGISTs. After controlling for baseline differences using PSM, the study systematically compared the perioperative and medium- to long-term oncological outcomes of the two surgical approaches. The principal finding was that for selected LGIST cases, endoscopic resection showed significant advantages, as assessed by perioperative indicators, while achieving comparable medium-term oncological outcomes to those of laparoscopic surgery. Specifically, the endoscopic group was significantly superior to the laparoscopic group in terms of operation time, intraoperative blood loss, and length of postoperative hospital stay (P < 0.05). This advantage remained consistent in the subgroup analyses stratified by tumor location (favorable/unfavorable sites) and growth pattern (intraluminal type). The advantages associated with the minimal invasiveness of endoscopic surgery were fully demonstrated: The use of transluminal surgery through natural orifices avoided trauma to the abdominal wall, thereby leading to faster postoperative recovery and shorter hospital stay, consistent with the conclusions of previous studies on endoscopic techniques[18]. The more rapid postoperative recovery of the endoscopic group is closely related to its unique minimally invasive characteristics. First, resection via natural orifices avoids the trauma of abdominal wall incisions, thereby reducing postoperative pain and the need for analgesics, and contributing to early mobilization. Second, the procedure does not interfere with the peritoneal cavity and intestinal serosa, minimizing disruption of intestinal motility and enabling earlier resumption of oral feeding. Third, the reduced surgical trauma is associated with less systemic inflammation, enabling the rapid recovery of organ function. These factors collectively contribute to the advantages of endoscopic resection in terms of postoperative hospital stay and functional recovery.

In terms of resection completeness, there were cases of incomplete resection in both groups. Zhao et al[19], in a retrospective analysis comparing endoscopic and surgical resection for gastric GISTs with diameters < 5 cm, reported an R1 resection rate in the endoscopic group of 4.7% (4/85). This was presumably related to unclear boundaries between the tumor and muscular layer tissue during endoscopic resection, as well as difficulties in the identification of tumor margins[14,18]. However, R1 resection has been found not to increase the risk of postoperative recurrence[20]. A study showed that endoscopically treated cases in which the tumor margins were difficult to identify due to piecemeal snare resection were more common when using immature early-stage endoscopic techniques, while en bloc resection could be achieved using EFTR[21]. Furthermore, cases of resection failure were observed to involve factors such as increased difficulty of endoscopic resection of exophytic lesions and limited operating space during tunnel resection[22]. It has also been observed that cases of resection failure using laparoscopy were associated with tumors located at sites unfavorable for laparoscopic manipulation[14]. In the present study, after balancing the sample size and baseline data between the two groups via PSM, there was no significant difference in the rate of complete resection between the endoscopic and laparoscopic groups, indicating that despite incomplete resection in a small number of cases in the endoscopic group, the difference in the technical success rate between the two groups was non-significant.

In terms of safety, the primary complications in the endoscopic group were perforation and infection. These complications are associated with increased risks of postoperative perforation and peritonitis due to incomplete closure of intraoperative perforations, prolonged dissection time, and larger wound surfaces. Notably, perforation in the endoscopic group was mostly the result of planned controlled perforations during EFTR, which form an inherent part of the surgical procedure and can be repaired effectively using mature endoscopic closure techniques. In contrast, unexpected perforations occurring during ESD are technical complications that require timely intervention. This distinction indicates that the safety profile of endoscopic resection should be interpreted in terms of the specific surgical technique, and planned controlled perforations should not be equated with unanticipated complications in risk weighting. Shedding of the tumor into the abdominal cavity during endoscopic resection has also been reported; this can be prevented by grasping the lesion with a snare or foreign-body forceps during EFTR[23]. The complications in the laparoscopic group were mainly infection and bleeding, potentially related to surgical trauma and stress[24]. Zhao et al[19] compared 152 cases of gastric GISTs with diameters < 5 cm treated by either endoscopic resection (85 cases) or open surgery (67 cases), and found that the incidence of intraoperative and postoperative complications in the endoscopic group was lower than that in the open-surgery group (5.9% vs 6.4%, P < 0.01), with intraoperative bleeding being the main complication in the latter (63.6%, 7/11). However, the tumor diameters in the endoscopic group were smaller than those in the open-surgery group (2.8 cm vs 4.3 cm, P < 0.05). A meta-analysis including 1165 cases comparing endoscopic and laparoscopic resection for gastric GISTs found that there was no significant difference in the incidence of complications between the two groups; however, the analysis had the problem of inconsistencies in tumor sizes between groups[25]. Zhu et al[26] compared 85 cases of gastric GISTs with diameters of 2-5 cm treated using endoscopic (43 cases) or laparoscopic (42 cases) resection, and observed a significantly higher rate of complications in the endoscopic group compared to the laparoscopic group (35.6% vs 0%, P < 0.01), with all complications in the endoscopic group being intraoperative and postoperative perforation. The discrepancy in complication rates between our study and that of Zhu et al[26] may be attributed to three key factors. First, differences in patient baseline characteristics: Zhu et al[26] included only GISTs located at the esophagogastric junction, which are known to have a high risk for complications, while our study covered a broader range of tumor locations, including favorable sites associated with fewer resection difficulties. Second, technical improvements: Our team prefers to use EFTR for tumors with transmural growth or serosal adhesion, thereby avoiding forced dissection that increases the risk of perforation, whereas Zhu et al[26] mainly used ESD for such cases. Third, perioperative management: Our adoption of the ERAS protocol, involving early mobilization and standardized infection prevention, may have reduced the incidence of postoperative complications. These differences indicate that the safety of endoscopic resection is closely related to patient selection, technical selection, and perioperative management, and that standardized procedures can effectively control complication risks. Since GISTs originate from the muscularis propria, intraoperative perforation is often unavoidable during endoscopic resection, but it can usually be effectively managed using endoscopic techniques without the need for additional intervention. Zhu et al’s inclusion of intraoperative perforation as a complication when comparing endoscopic and laparoscopic resection would have tended to overestimate complications in the endoscopic group[26]. After standardization of the baseline data between the two groups with PSM, no significant difference was found in the complication rate, suggesting that despite the occurrence of perforation and infection in a small number of patients in the endoscopic group, there was no statistically significant difference between the safety of the two procedures.

In terms of perioperative indicators, the findings of Zhao et al[19] are consistent with those of the present study, showing that the endoscopic group had lower hospitalization costs, earlier resumption of oral feeding, and shorter postoperative hospital stay. The present study further confirmed that for selected patients with gastric GIST, endoscopic resection had significant advantages over laparoscopic resection in terms of perioperative indicators (operation time, intraoperative blood loss, postoperative hospital stays; P < 0.05), and achieved comparable medium-term oncological outcomes. Notably, despite the higher technical difficulty of endoscopic resection, its oncological radicality was not compromised, as seen in the comparable rates of R0 resection and intraoperative tumor rupture between the two groups, confirming that in high-level medical centers, endoscopic techniques also adhere to the principle of tumor-free surgery. After a median follow-up of 32 months, no significant differences were observed in the RFS and OS rates between the two groups, while multivariate Cox analysis confirmed that tumor size, mitotic count, and NIH risk stratification were independent predictors of prognosis, although surgical approach was not predictive. This suggests that after achieving R0 resection, the choice between endoscopic and laparoscopic surgery does not influence the risk of recurrence determined by the inherent biological characteristics of the tumor.

The results of this study have important clinical implications in showing that endoscopic resection represents a feasible alternative to laparoscopic surgery for the treatment of selected patients with gastric GIST. For intermediate- to low-risk tumors ≤ 3 cm in diameter located at favorable sites, such as the gastric fundus and gastric body, endoscopic surgery may be the preferred option due to its minimal invasiveness. For tumors located at unfavorable sites, such as the gastric antrum and cardia, although the subgroup analysis showed that endoscopic resection also had significant advantages, due to the high level of technical difficulty, it should be performed in experienced medical centers after careful evaluation by a MDT. The MDT evaluation should include a comprehensive assessment of the anatomical characteristics of the tumor, the technical feasibility of endoscopic resection, and patient preferences, to maximize benefits.

This study has several limitations. First, it was a single-center retrospective analysis. Although PSM was used to reduce bias, selection bias, and the presence of confounding factors cannot be avoided completely. Second, while the sample size was sufficient to detect differences in perioperative indicators, there was insufficient statistical power to assess low-probability events such as long-term recurrence and survival. This thus represents an important interim report on the mid-term outcomes of endoscopic and laparoscopic resection for LGISTs. While the median follow-up of 32 months provides reliable evidence for short-to-medium-term safety and efficacy, long-term recurrence (e.g., five years or longer) remains to be verified. Our research team is currently conducting an extended follow-up of the cohort, with a planned duration of 5-8 years to confirm the long-term oncological equivalence of the two surgical approaches. For intermediate-risk patients, in particular, prolonged follow-up could clarify whether endoscopic resection can maintain comparable RFS rates to those of laparoscopic resection beyond three years. Longer follow-up times and larger sample sizes may be needed to confirm the long-term oncological equivalence of the two surgical approaches. Third, endoscopic surgery is associated with a steep learning curve, and the results of this study may therefore reflect outcomes obtainable at high-level medical centers, requiring verification of generalizability in a wider range of institutions. In the future, multicenter, prospective, randomized controlled trials are needed to provide higher-level evidence. In addition, longer follow-up (five years or more) is crucial for evaluating the long-term efficacy of the two surgical approaches. Preoperative imaging radiomics models based on artificial intelligence may provide greater accuracy in predicting tumor behavior, thereby providing support for the individualized selection of the optimal surgical approach.

CONCLUSION

The results of this study indicate that endoscopic resection is a safe and effective minimally invasive option for the treatment of carefully selected patients with LGISTs. While preserving the oncological efficacy of laparoscopic surgery, endoscopic resection offers significant advantages in reducing surgical trauma and accelerating patient recovery. The selection of surgical approach should be individualized, based on the anatomical characteristics of the tumor, MDT assessment, and the condition of the patient, rather than simply following traditional practices. The findings of this study support the application of endoscopic techniques to the treatment of gastric GISTs.