Stratified and propensity score matching analysis of the effectiveness of fibrin glue in preventing postoperative bleeding after endoscopic submucosal dissection.

1.
Introduction
Endoscopic submucosal dissection (ESD) is a sophisticated technique that overcomes the limitations of conventional endoscopic mucosal resection (EMR). It offers the benefits of a less invasive procedure and a low postoperative local recurrence rate, which is now a standard treatment for upper gastrointestinal lesions (including esophageal and gastric superficial tumor lesions) [1–3]. However, ESD can cause complications such as perforation, bleeding, and esophageal stenosis depending on the equipment, technology, operator experience, and the patient’s overall health and postoperative bleeding is a common issue of ESD. Severe cases might end in hemorrhagic shock, requiring initial blood transfusion, interventional radiology, or even surgery. Specifically, the bleeding rate can be as high as 4.1% to 8.5% in patients with early gastric cancer (EGC) [4,5]. Antithrombotic drugs and large resection areas increase the danger of bleeding after ESD and in people with these characteristics, the bleeding rate after ESD might range from 21 to 38% [6]. Furthermore, the possibility of delayed bleeding and rebleeding after ESD is substantially greater in high-risk individuals who take warfarin and have regular dialysis [5,7]. Therefore, it is of great significance to explore effective methods to prevent postoperative hemorrhage following upper gastrointestinal ESD.
Fibrin glue (FG) is a tissue adhesive and hemostatic substance, mainly composed of human or bovine blood fibrinogen and thrombin [8], which are frequently utilized in neurosurgery, ophthalmology, and other specialties [8–10]. Previous research explored the risk of delayed bleeding following ESD by covering the wound with a polyglycolic acid (PGA) sheet and FG [11–13]. However, the cost of PGA sheets mixed with FG is exorbitant, and the placement and fixation methods during surgery are technically challenging and laborious, restricting their application. Wang et al. demonstrated that in patients with stomach precancerous lesions, only porcine FG was employed as a hemostatic method during ESD surgery, enhancing the healing of ESD-induced ulcers without increasing the level of delayed bleeding [14].
In the prevention of post-ESD bleeding, several critical knowledge gaps remain. Currently, there is minimal data on the utilization of FG independently for decreasing bleeding after ESD, and these studies focus primarily on superficial lesions such as carcinoma, precancerous lesions, and adenomas [14,15]. The impact of FG spraying on postoperative bleeding in patients with early cancers has not been sufficiently explored. Moreover, the comprehensive influence of high-risk factors, such as the use of antithrombotic drugs, lesion location, and size, on postoperative bleeding requires further clarification. Thus, this study aims to explore the impact of FG spraying on postoperative bleeding in patients with EGC and early esophageal cancer (EEC). Additionally, it explores potential risk factors for bleeding after upper gastrointestinal ESD, offering extra options for ESD evaluation and therapy, along with perioperative management of patients with early upper gastrointestinal cancer.

2.
Methods
2.1.
Patient selection
This study included patients with early cancer (including esophageal and gastric early cancer) who underwent upper gastrointestinal ESD in the First Affiliated Hospital of Nanchang University from November 2022 to November 2024, and retrospectively analyzed resources, including clinical records, endoscopic surgery reports or monitoring, and telephone follow-up data. Patients with high-grade intraepithelial neoplasia (HGIN) and carcinoma confirmed by postoperative pathology were included. The exclusion criteria were: (1) Incomplete clinical data; (2) Remnant stomach; (3) Perforation or partial full-thickness resection during ESD.
According to standard guidelines [16,17], all patients taking antiplatelet drugs such as aspirin or clopidogrel were asked to stop these drugs at least 5 days before ESD. The drugs were resumed on days 3–5 after ESD. Oral warfarin was stopped 5 days before surgery, and new oral anticoagulants like rivaroxaban were stopped one day prior to surgery. For patients taking oral anticoagulants, low molecular weight heparin bridging therapy should be applied in the medication-free period, and the use of anticoagulants should be resumed within 24 h after the operation, per the guidelines.
Patients were separated into two groups: the fibrin glue group (FG group) and the conventional group (Figure 1).
All subjects signed informed consent for ESD, and in the FG group, when the operating doctor decides to use FG during the operation, the legally authorized client of the patient would be notified immediately and signed informed consent for the use of FG. The study complied with the Declaration of Helsinki and was approved by the Ethics Committee of the First Affiliated Hospital of Nanchang University, with the approval number 2025-338.

2.2.
ESD procedure and application of FG
In principle, ESD was performed according to standard methods using a therapeutic endoscope (GIF-Q260J; Olympus Medical, Tokyo, Japan) in the Gastrointestinal Endoscopy Unit of the First Affiliated Hospital of Nanchang University. The steps mentioned below were taken: 1) ring marking 3–5 mm from the lesion border; 2) injecting 0.2% to 0.4% sodium hyaluronate solution or saline submucosally around the lesion in a proximity-to-distance order; 3) cutting the mucosa and removing the submucosal layer with the ESD device; 4) using hemostatic forceps to coagulate the visible blood vessels at the base of the ESD ulcer. All ESD wounds were pre-coagulated by traditional hemostasis methods such as electrocoagulation and hemostatic clips under the judgment of endoscopists when it is necessary. The aforementioned hemostasis techniques will also be used to stop intraoperative bleeding in order to avoid interfering with subsequent steps. For the FG group, FG gel was sprayed after this step to ensure no active bleeding. Subsequently, the post-ESD wounds were uniformly covered with porcine FG (BIOSEAL, Guangzhou Bexiu Biotechnology Co., Ltd., China) via endoscopic spray (Figure 2). FG uses a fixed dose, which is based on completely covering the wound surface of ESD. The gel fully solidified on the ESD wound surface five minutes after the fibrin glue was sprayed. Minor oozing persisting may occasionally be addressed with FG adjunctively, but this should not substitute primary hemostatic interventions. All procedures were performed by several endoscopists who had performed more than 200 ESDs and were trained and experienced in diagnostic and therapeutic procedures under the supervision of ESDs.

2.3.
Perioperative management and follow-up
All included patients exercised preoperative magnifying chromoendoscopy and ultrasound endoscopy. Fasting and water restriction for 8 h before surgery. Fasting for 48 h started after surgery. A standard dose of intravenous proton pump inhibitor (PPI) was administered intravenously within 24 h of ESD for 2 days, followed by oral PPI (40 mg, once daily). Gastric ESD was taken orally for 8 weeks, while esophageal ESD was given orally for 4 weeks. Hematemesis, melena, or a drop in hemoglobin of ≥2.0 g/dL is considered postoperative bleeding, and the attending physician will send the patient right away for emergency endoscopic surgery. Prior research has demonstrated that routine secondary endoscopy is not essential for preventing prolonged bleeding following ESD [18–21], and secondary endoscopy was not routinely performed in this study.

2.4.
Data collection
The necessary baseline information was obtained by querying the case and endoscopic procedure systems. Bleeding events were gathered by reviewing case records, emergency endoscopy records, and phone follow-ups. The patient’s gender, age, history of anticoagulant use, history of smoking, information on alcohol consumption, and concomitant diseases (hypertension, diabetes mellitus, cardiovascular and cerebrovascular disease, cirrhosis of the liver, chronic kidney disease, and hyperlipidemia) were among the general baseline and clinical characteristics that were gathered. Additionally included were pathologic information like lesion size, location, negative pathologic margins, en bloc resection, and complete resection. In the case of multiple lesions, the maximum diameter of the largest lesion was taken. The location of gastric lesions is categorized into upper (U), middle (M), and lower (L) according to the classification of gastric cancer by the Japanese Gastric Cancer Association [22]. Additionally, surgery-related data such as preoperative PPI use, intraoperative hemostatic clip use, the presence or absence of intraoperative bleeding, as well as laboratory indicators such as Hb, PLT, APTT, and PT-INR were recorded. All biochemicals were collected within 24 h preoperatively. Preoperative proton pump inhibitor (PPI) use was understood as two standard-dose intravenous injections of PPI 24 h prior to surgery.

2.5.
Histological evaluation
After the procedure, the lesion mucosa was recovered and unfolded, the needle was fixed on a foam plate along the outside of the marking, the proximal and distal ends of the lesion were marked, the maximum diameter of the lesion was measured, the specimen was fixed in 10% formalin solution, and serial sections of the specimen were made at 2 mm intervals.

2.6.
Definitions and clinical outcome
Delayed bleeding events were defined as the occurrence of hematemesis or melena, a decrease in Hb level >2 g/dL, and bleeding confirmed by endoscopy. The primary outcome measure was the overall bleeding rate (OB) within 4 weeks after ESD. We will concentrate on bleeding events within 48 h after ESD and after 48 h in both groups to better examine the impact of FG on bleeding following ESD, as the number of bleeding events within 24 h after surgery is still appropriate [15,23]. Thus, the acute bleeding rate within 48 h after surgery (AB) and the delayed bleeding rate beyond 48 h (DB) are the secondary observation indicators. In addition, the intraoperative bleeding rate (IB) and adverse events, such as an allergic reaction to protein glue, were noted. Intraoperative bleeding events were defined as: 1) multiple use of electrotome, hemostatic forceps, or at least use of hemostatic clips can effectively control bleeding; 2) active bleeding during treatment was significant enough to need operation interruption and/or a blood transfusion [23,24].

2.7.
Statistics
Normally distributed quantitative data were expressed as mean ± standard deviation, and p-values were calculated using a t-test. Skewed quantitative data were expressed as the median and interquartile range (IQR), and p-values were calculated using the Mann–Whitney U test. Qualitative data were expressed as rates and constituent ratios, and p-values were calculated using the χ2 test or Fisher’s exact test. The bleeding rates between groups were compared using the χ2 test or Fisher’s exact test.
Firstly, we compared the baseline of the FG group and routine group with SPSS software and analyzed the postoperative bleeding rate of EGC and EEC in the intraoperative bleeding group and non-intraoperative bleeding group by a stratified method. Secondly, to eliminate bias due to confounding variables, the propensity score matching (PSM) method was utilized, and the FG and conventional groups were matched 1:1, with a matching score difference of less than 0.02. Additionally, univariate and multivariate logistic regression were used to explore the risk factors associated with delayed bleeding after ESD. We also used a tool to compute statistical power analyses, G Power software (Version 3.1.9.7, Franz Faul, Kiel University, Kiel, Germany) (α = 0.05, a power of 0.90), to calculate and verify that the total sample size had reached the proper statistical level.
SPSS software, version 27.0 (IBM Corp., Armonk, N.Y., USA) and R software (version 4.4.2, R Core Team, Vienna, Austria) were used for statistical analysis, and p < 0.05 was considered statistically significant.

3.
Results
3.1.
Baseline characteristics
Out of 1340 patients who received upper gastrointestinal tract ESD treatment, we chose 362 patients with early-stage cancer and split them into two groups: the conventional group (n = 211) and the FG group (n = 151) (Figure 1). There was no significant difference between the two groups patients’ gender, smoking, alcohol use, antithrombotics, comorbidities, anticoagulant medication use, lesion location, histologic type, intraoperative hemostatic clip use, complete resection, and En bloc resection. Both teams’ laboratory indices fell within normal ranges. Patients in the FG group were 66 ± 8.0 years older on average than those in the conventional group, which was 62.0 ± 9.0 years old with a larger size of lesion resection. Patients in the FG group experienced a significantly higher intraoperative bleeding rate (76.8%) than those in the non-glue group (37.9%), p < 0.001. Furthermore, the glue-sprayed group had a statistically significantly higher percentage of SM (9.9%) in the depth of lesions than the non-glue-sprayed group (3.3%) (Table 1).

3.2.
Postoperative bleeding rates in patients with EGC or EEC
Overall, the stomach had a higher postoperative bleeding rate than the esophagus, with the OB, AB, and DB rates of 8.3%, 4.0%, and 4.3% for the stomach, and 1.6%, 3.2%, and 4.8% for the esophagus, respectively (Table 2). There was no statistical difference in the above bleeding indicators between the esophagus and the stomach. Moreover, there is no significant statistical difference in bleeding rate after ESD before matching between the FG group and the control group (Tables 3–5).

3.3.
Comparing various types of EGC and EEC postoperative bleeding in a stratified method
Given that the intraoperative bleeding rate appeared to be significantly different between the FG and conventional groups, the study population of 362 cases was stratified based on whether or not they were IB, and the bleeding events in the stomach and esophagus were further investigated (Table 6). In the intraoperative bleeding group, 8 (9.0%) patients who underwent gastric ESD with FG developed OB, while the group that underwent gastric ESD without FG discovered 13 (18.8%) cases of OB (p = 0.07), which was not statistically significant. The proportion of EGC patients who received FG who developed AB was 2.2%, which was considerably lower than the 11.6% of patients in the same group who did not receive FG (p = 0.039), indicating a significant statistical difference and demonstrating that intraoperative spraying of FG in patients with EGC reduces the incidence of bleeding events in the 48-hour postoperative period. For EEC, there was no statistical difference in postoperative bleeding rate between the FG group and routine group, regardless of stratification or not (p > 0.05).

3.4.
Baseline attributes following the matching of propensity scores
Under the condition of the original grouping, the age, lesion size, and depth of the two groups were statistically different (Table 1). In this study, the PSM method was utilized to exclude the samples affecting the baseline balance to analyze the effect of the use of FG on delayed hemorrhage in upper gastrointestinal ESD in a more rational way. To establish new baseline characteristics for the 1:1 matched FG group (n = 102) versus the conventional group (n = 102), all baseline metrics were incorporated into the PSM (Table 7). The baseline variables such as age (p = 0.934), gender (p = 0.763), anticoagulants (p > 0.999), intraoperative bleeding (p = 0.770), lesion size (p = 0.748), and lesion depth (p > 0.999) did not differ by a significant percentage between the two novel matched groups.

3.5.
Analysis of bleeding events between the FG group and the conventional group after PSM
Based on the baseline data obtained after PSM, we calculated the incidence of bleeding events in both of the novel groups (Table 8). The OB rate of 102 patients with early-stage cancer who used FG was 2.9%, which was significantly lower than that of the conventional group of 12.7% (p = 0.009). Similarly, the superiority of having used FG was equally evident in the analysis of AB events, with a rate of 1.0% in the FG group vs. 7.8% in the conventional group, respectively (p = 0.035). However, in both groups, the incidence of DB was not statistically significant (p = 0.445). The aforementioned findings were consistent with the stratified analysis conducted before PSM, which also revealed that patients with early-stage upper gastrointestinal tract malignancies could experience a reduced risk of postoperative bleeding following ESD if FG was used wisely.

3.6.
Independent risk factor screening for overall delayed bleeding (OB) events
In this study, using pre-matched baseline information, we included variables such as age, lesion depth, comorbidities (cardiovascular disease, diabetes mellitus), negative pathologic margins, intraoperative bleeding, intraoperative use of hemostatic clips, preoperative use of PPIs, and anticoagulant use in one-way regression analyses associated with the presence of OB. The findings of intraoperative bleeding (OR = 4.28, 95% CI 1.59–11.53) and negative pathologic margins (OR = 3.54, 95% CI 1.09–11.52) were included in the multifactorial logistic regression analysis since they were statistically significant (Table 9). In patients with early-stage upper gastrointestinal cancers, the only independent risk factor for postoperative bleeding following ESD was intraoperative bleeding (OR = 4.00, 95% CI 1.47–10.84), which indicates that the incidence of intraoperative bleeding might affect the outcome of postoperative bleeding.

4.
Discussion
Delayed postoperative bleeding, the most common complication of gastric ESD, has always drawn a lot of attention, and there is no shortage of studies on it around the world [25–28]. FG has been used extensively in surgery for hemostasis, suture support, and tissue adhesions. The FG spraying on wounds is relatively easy to operate and reasonably priced in the ESD. Additionally, FG is also quite resistant to corrosion at low PH [9,13,15]. However, studies that solely used FG as a hemostatic agent to prevent bleeding were essential to be further conducted. Compared with earlier research, this study focuses on the early cancer population of the gastric and esophageal and explores the effects of FG alone on postoperative bleeding. After stratified comparisons of bleeding rates between the groups with and without FG, we compared bleeding after ESD using PSM, a statistical method to rule out potential confounding factors.
In this study, a total of 362 patients with early upper gastrointestinal tract cancers including 300 cases of EGC and 62 cases of EEC were enrolled. Four patients who required a three-arm titanium clip and purse-string suture owing to intraoperative perforation or partial full-thickness resection were not enrolled in our EGC patient recruitment for the reason that the extra procedure greatly extended the period of the procedure and possibly had an impact on FG’s capability to prevent post-ESD bleeding episodes [5,29,30]. Moreover, only one patient confronted a bleeding rash following intraoperative FG application, which went away on its own a day later, and none of the patients treated with ESD experienced postoperative perforation.
In patients with EGC, the bleeding rate following ESD was 25/300 (8.3%), which is consistent with the previous study that the bleeding rate of patients with early gastric cancer after ESD is 4.1%-8.5% [7,31,32]. According to earlier research, the postoperative bleeding rate following ESD for esophageal cancer can range from 2% to 4% [33–36], which is nearly close to our findings (4.8%). Moreover, the bleeding rate after EGC compares with that after EEC, with no significant statistical difference (Table 2), giving the subsequent statistical analysis a theoretical foundation. To further analyze the effect of FG on the bleeding rate of the esophagus and stomach after ESD, we also carried out a stratified analysis. In addition, previous studies have shown that the bleeding rate after stomach ESD is higher than that after esophageal, which is related to the fact that there is no gastric acid stimulation in the esophagus and the influence of postoperative artificial ulcer is reduced on the grounds that esophageal peristalsis is far less than that in the gastrointestinal tract [4,36,37].
After PSM analysis, even based on preventive hemostasis measures, the postoperative bleeding rate of the routine group is 12.7%, which may be as follows: As shown in Table 1, the use of FG decided by the endoscopists is biased according to the intraoperative conditions such as the size and depth of the lesion, and intraoperative bleeding, by matching with the control group, the samples with high postoperative bleeding events are inclined to be retained after PSM analysis.
Currently, the methods that have been determined to be effective in preventing delayed bleeding after ESD include continuous use of PPIs after surgery and coagulation or clipping of visible blood vessels in the ulcer [38,39]. However, even with these preventive methods, the incidence of post-ESD hemorrhage is about 5% [29]. Shiratori et al. found that the use of the Doppler probe method (DOP) may be an effective means of reducing the incidence of delayed hemorrhage in ESD [40]. Hemostatic powder or gel is also used as a clinical method to prevent hemorrhage in some areas. The principle is to prevent delayed bleeding by accelerating ulcer healing [29]. Plus, some biodegradable polymers have emerged as a new area of interest for post-ESD bleeding prevention research in recent years. PGA sheets, one of the earliest biodegradable polymers investigated for biomedical uses, have been attempted to stop bleeding after ESD. The combination of PGA and fibrin sealant has now proved to be a powerful adhesive and is now commonly used in surgical procedures [29,41,42]. Kawata et al. showed that PGA tablets in combination with FG in ESD can prevent postoperative bleeding by covering the surface of artificial ulcers in a population at high risk of bleeding on anticoagulants [12]. It is worth noting that a prospective RCT trial verified that in patients at high risk of bleeding, applying FG to ESD-induced gastric ulcers could not effectively reduce bleeding within 4 weeks after ESD, but it could close iatrogenic ulcers in the acute phase, thereby reducing the occurrence of delayed bleeding [15]. However, instead of focusing on patients with early cancer, the RCT is based on people with stomach tumors and high-risk bleeding. Our study included all accessible cases in the First Affiliated Hospital of Nanchang University from 2022 to 2024. Strict inclusion and exclusion criteria were used to obtain a total of 362 samples, all while adhering to the applicable retrospective study guidelines [43]. Despite the sample size’s limitations, it offers effective parameters for the prospective investigation of bleeding following ESD in patients with early-stage upper digestive tract cancer.
In this study, by forming a retrospective cohort, after comparing the baseline, it was found that in this item of intraoperative bleeding, there was a significant difference between the sprayed group (76.8%) and the conventional group (37.9%) (p < 0.001). Thus, stratification based on intraoperative bleeding was performed to explore the difference in postoperative bleeding rates between the gel-sprayed group and the conventional group in terms of stomach and esophagus (Table 6). The results demonstrated that only in the case of intraoperative bleeding, the use of FG in patients with EGC undergoing ESD had a preventive effect on the 48-h postoperative bleeding rate. In other words, FG reduces postoperative bleeding events in the short term after ESD for EGC, which is consistent with the findings of Lee et al. [15]. In contrast, in esophageal ESD, spraying or not spraying FG had no significant effect on the postoperative bleeding rate. This discrepancy in efficacy between gastric and esophageal ESD wounds may be attributed to two principal factors. First, PSM analysis was employed to mitigate selection bias, resulting in a predominantly gastric ESD cohort following the exclusion of unmatched cases. Second, the limited sample size of EEC and their inherently lower bleeding incidence after esophageal ESD relative to gastric ESD likely reduced statistical power to detect clinically meaningful differences, thereby increasing the negative rate in outcome interpretation. Furthermore, there was no significant difference in antithrombotic use between the two groups that the distribution of aspirin, clopidogrel, and DOACs showed no statistical significance (Table 1). Although antithrombotics are established risk factors for postoperative bleeding [6], their potentially confounding effects in this cohort may have been minimized due to low overall usage rates and strict perioperative management.
Although the original baseline is basically the same, variables including the age, lesion size and lesion depth between the conventional group and the FG group are statistically significant, which has been indicated to be likely linked to post-ESD bleeding by previous reports [23,24,44]. To make the analyzed data more equivalent to what prospective randomized data sets would reveal, PSM is a statistical tool that straightens out known imbalances by selecting patients with consistent background information, without recollecting patient information. We included all baseline indicators into PSM for 1:1 matching to obtain the new FG group and the conventional group (Table 8). The robust post-PSM analysis strengthens causal inference and supports FG’s clinical utility for acute bleeding prevention. The baseline of the new two groups was highly consistent, and further comparison of the OB, AB, and DB of the two groups showed that, consistent with the stratification analysis, FG had a delaying effect on postoperative acute bleeding after ESD for early upper gastrointestinal cancer. Perhaps consistent with the effect of PGA tablets, for bleeding in the short term after ESD, the spray adhesive can prevent bleeding by covering the wound to prevent bleeding caused by post-ESD ulceration [11–13]. However, for delayed bleeding beyond 48 h, the use of FG is obviously of little significance (Table 9). The method of preventing bleeding after ESD may be to promote hemostasis through clot formation without producing foreign body reactions and to promote wound healing through tissue sealing. In addition, compared with PGA sheets, spraying FG is relatively simple and less expensive, and more prospective and multicenter studies are expected to be added to further explore the application of FG in upper gastrointestinal ESD.
Univariate and multivariate regression analysis was used to further explore the variables associated with postoperative bleeding after ESD for early upper gastrointestinal cancer. Intraoperative bleeding and negative horizontal and vertical resection margins emerged as significant predictors in univariate analysis. Multivariate analysis further confirmed intraoperative bleeding as an independent risk factor, aligning with prior studies suggesting its role in influencing postoperative bleeding following ESD [23,24,45]. This could suggest that by treating intraoperative bleeding as a stratified variable in our study, the impact of confounding variables on intraoperative fibrin glue spraying on postoperative bleeding will be significantly diminished. Therefore, we draw Table 6, which implies that spraying fibrin glue can prevent postoperative bleeding in certain circumstances. Notably, the study by Liu et al. showed that intraoperative bleeding during ESD was not related to delayed bleeding. However, the study targeted gastrointestinal stromal tumors and only counted patients with severe intraoperative bleeding, ignoring a large number of patients with minor intraoperative bleeding [46]. The causes of intraoperative bleeding’s influence on postoperative bleeding were investigated in this study: Intraoperative bleeding can play a crucial role in the operation’s outcome, lengthening its duration and increasing the risk of postoperative bleeding following ESD [30,32, 46,47].
It is worth mentioning that the preoperative use of PPIs was not associated with postoperative bleeding in ESD in our cohort, which may be related to the fact that it was administered by the intravenous route preoperatively and lasted for only 1 day. This null effect may be attributed to the abbreviated therapeutic duration and route of delivery, as evidence suggests that preoperative oral administration of PPIs for 1 week has a greater effect on intraoperative PH, which is beneficial in controlling the occurrence of hemorrhage [48–50]. Furthermore, the potential impact of gastric acid on FG warrants further investigation. Although the values and changes of gastric acid were not systematically analyzed, the corrosive effect of gastric acid on FG is likely minimal in this context. To suppress acid secretion during and after the procedure, all patients received standardized intravenous PPI therapy. This regimen aligns with evidence demonstrating that FG maintains structural integrity for 168 h and provides effective wound protection exceeding 7 days when the pH value of pepsin is 6.0 [9,13,15]. Given that PPI therapy elevates intragastric pH above 4.0 within 24 h, the combined protocol likely preserves FG’s mechanical barrier function during the critical 48-hour post-ESD period, thereby reducing early bleeding risks.
Nevertheless, this study has several conspicuous limitations inherent to its retrospective design. First, despite PSM to balance measured confounders like lesion size, and intraoperative bleeding, unmeasured variables such as subtle endoscopic technique variations or transient hemodynamic changes during ESD may have influenced outcomes and produced the bias. Second, the single-center design limits generalizability, as all procedures were performed by skilled endoscopists, potentially underestimating bleeding risks in less experienced settings. Third, the small sample size of esophageal ESD cases precludes definitive conclusions on FG’s efficacy in this subgroup, warranting validation in larger cohorts. While our findings suggest FG reduces acute bleeding in EGC, RCTs are essential to confirm causality and optimize FG application protocols. Future multicenter RCTs should prioritize patients with EGC and standardize FG dosing, spray techniques, and postoperative acid suppression regimens. Additionally, mechanistic studies exploring FG degradation kinetics relative to ulcer healing phases could clarify its time-limited efficacy.

5.
Conclusion
Fibrin glue sprayed intraoperatively may help reduce bleeding following ESD, particularly in EGC patients who contribute to acute postoperative hemorrhage. In light of its correlation with postoperative bleeding, intraoperative bleeding in ESD should be valued in procedures.