The efficacy and safety of endoscopic ultrasound selective variceal devascularization for the treatment of esophagogastric varices: a retrospective cohort study

Abstract

Background

Gastroesophageal variceal bleeding is a life-threatening complication of portal hypertension. This study evaluated the efficacy and safety of endoscopic ultrasound-guided selective variceal devascularization (EUS-SVD) compared with conventional endoscopic cyanoacrylate injection.

Methods

This retrospective cohort study adhered to the STROBE guidelines. Between August 2022 and June 2024, 107 patients were enrolled and allocated to conventional endoscopy (n = 52) or EUS-SVD (n = 55). Propensity score matching was performed to reduce confounding. Primary outcomes included rebleeding, ulcer formation, and resource utilization.

Results

Compared with the conventional group, the EUS group required significantly lower volumes of polidocanol and cyanoacrylate (both P < 0.05). The 6‑month rebleeding rate and 3‑month ulcer formation rate were significantly lower in the EUS group (P < 0.05). Kaplan-Meier analysis demonstrated a reduced cumulative rebleeding incidence in the EUS group (HR 0.38, 95%CI 0.17–0.86, P = 0.021).

Conclusions

Within the limitations of a single-center retrospective design, these findings suggest that EUS-SVD may represent a promising therapeutic option for gastroesophageal varices, although it requires validation in prospective randomized trials.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12876-026-05151-3.

Background

Liver cirrhosis is a common liver disease in clinical gastroenterology. When cirrhosis progresses to the decompensated stage, collateral circulation forms, leading to portal hypertension (PH), which results in gastroesophageal varices [1] (GOVs). Gastrointestinal bleeding, presenting as melena, hematemesis, and hemorrhagic shock, is a clinical manifestation of variceal rupture [2] and one of the main causes of death in patients with cirrhosis [3, 4]. Therefore, taking timely hemostatic measures and which prevents rebleeding are key to reducing mortality in these patients [5]. Currently, clinical treatments for GOV include surgical procedures, interventional therapy, medication, and gastrointestinal endoscopy techniques. Both surgical approaches, such as splenectomy and devascularization, and interventional embolization focus on addressing the source of the portal-systemic shunts. In contrast, endoscopic treatments intervene at the peripheral side of the collateral circulation. While theoretically and practically, surgery and radiological interventions are more effective in preventing rebleeding than endoscopic treatments are, they are usually performed electively, and the high preoperative bleeding rate further burdens the liver. In recent years, with the increasing maturity of endoscopic ultrasound (EUS) in clinical practice, it has increasingly demonstrated its unique advantages in the prevention and treatment of esophageal and gastric varices [6, 7]. EUS enables precise localization and direct action on the key blood vessels causing varices. This targeted treatment strategy not only improves the cure rate but also, through refined management and control, effectively avoids the side effects associated with excessive use of cyanoacrylate glue, such as ectopic embolism, which can occur with traditional endoscopic treatments [8, 9]. This comprehensive approach enhances both treatment safety and patient quality of life. In the present study, we propose the term “endoscopic ultrasound‑selective variceal devascularization (EUS‑SVD)” to describe a refined EUS‑guided therapeutic strategy. Unlike conventional EUS‑guided cyanoacrylate injection, which primarily targets the varix lumen itself, EUS‑SVD emphasizes the identification and obliteration of the feeding and draining collaterals responsible for varix perfusion. Under Doppler guidance, the puncture needle is selectively introduced into these key vessels rather than blindly into the varix cavity, aiming to achieve true devascularization—i.e., cessation of flow within the entire collateral network—rather than partial luminal filling. This conceptual shift prioritizes hemodynamic control over simple mechanical occlusion. Therefore, the purpose of this study was to analyze the efficacy and safety of EUS treatment compared with conventional endoscopic treatment in patients with GOV to assess the clinical value of EUS in managing gastroesophageal varices.

Methods

Study design and participants

A total of 107 patients with gastroesophageal varices (GOVs) who underwent elective endoscopic cyanoacrylate glue injection at the Gastrointestinal Endoscopy Center of Nantong University Affiliated Hospital between August 2022 and June 2024 were selected for the study. The patients were divided into two groups: a conventional endoscopy group (52 patients) and an EUS group (55 patients). During the perioperative period and follow-up, all patients underwent routine pulse oximetry monitoring. Patients with suspected embolism underwent chest and abdominal CT scans for confirmation. No symptomatic or asymptomatic ectopic embolism occurred in this study. Propensity score matching (PSM) was performed to reduce confounding biases.

The inclusion criteria for patients were as follows: (1) aged ≥ 18 years; (2) met the diagnostic criteria for moderate to severe GOV with a history of rupture and bleeding; (3) were eligible for endoscopic examination and cyanoacrylate glue injection treatment; and (4) provided signed informed consent. The exclusion criteria were as follows: (1) isolated gastric varices (IGV type); (2) absolute contraindications for endoscopic examination; and (3) concurrent malignant tumors such as liver cancer or portal vein thrombosis, severe cardiovascular diseases, or failure of other organs; (4) Patients who have had esophageal or gastric surgery (Fig. 1).

Fig. 1.

Flowchart of the different methods for treating esophagogastric varices: conventional endoscopy group direct endoscopic injection of cyanoacrylate; and EUS group, endoscopic ultrasonography-guided cyanoacrylate injection

Allocation Criteria: Since this was a retrospective study without randomization, the allocation to conventional endoscopy or EUS-guided treatment was determined by clinical practice conditions. Patients were selected for EUS treatment if the operator deemed the vascular anatomy complex or if resources permitted. Conversely, patients were treated conventionally if EUS equipment or expertise was unavailable at the time of admission. To minimize confounding bias, Propensity Score Matching (PSM) was subsequently employed.

Interventions

EUS Group: In this group, polidocanol and cyanoacrylate glue injections were performed with the assistance of endoscopic ultrasound (EUS). The procedure is as follows: (1) The conventional endoscope is inserted into the stomach to observe the mucosa of the lower esophagus and the gastric fundus near the cardia for varices (Fig. 2); (2) if varices are observed, it is switched to an ultrasound endoscope (EUS model: OLYMPUS GF-UCT260) for sector scanning (varices appear as echo-free areas on EUS, and Doppler flow signals can be detected), and the diameter and source vessels of the varices can be measured (Fig. 3); (3) under the guidance of Doppler ultrasound (ultrasound machine model: OLYMPUS EUME2 PREMIER PLUS), the puncture needle (Boston Scientific 22G endoscopic ultrasound biopsy needle) was selectively advanced into the main feeding or draining collateral vessels rather than solely into the varix lumen.; (4) guided by the EUS, injects polidocanol and cyanoacrylate glue [Surgical glue, N-butyl cyanoacrylate GLUBRAN2] via the “sandwich” technique (sugar-glue-sugar injection). Special attention was paid to ensuring that the polidocanol and cyanoacrylate glue reached the proximal and distal segments of the targeted collateral; (5) Successful devascularization was defined by complete loss of Doppler flow signals within the treated varix and its supplying vessels, EUS guidance allows observation of the echo-free area completely filled with high or mixed echoes; (6) after the injection, use color Doppler again to observe the blood flow in the varices and evaluate the embolization effect (Fig. 4).

Fig. 2.

Gastric fundus Tumor-like gastric fundal varices observed under conventional endoscopy visible under conventional endoscopy

Fig. 3.

A (upper right corner) EUS guided puncture of the feeder vessel using 22G needle (red arrow). B Varicose vein clusters appear hypoechoic under EUS and rich Doppler blood flow signals can be detected

Fig. 4.

Under EUS observation, the varicose vein mass after cyanoacrylate injection showed hyperechoic changes. The blood flow in the varicose vein mass significantly decreased after the injection of cyanoacrylate

Conventional Endoscopy Group: In this group, polidocanol and cyanoacrylate glue injections were performed with the assistance of a conventional endoscope. The procedure was as follows: (1) The conventional endoscope was inserted into the stomach; (2) the extent of the lesion was determined, and polidocanol and cyanoacrylate glue were injected via the “sandwich” technique (sugar-glue-sugar); and (3) After the injection, the embolization of the varices was performed under the guidance of the conventional endoscope.

Scheduled follow-up endoscopies were performed at 3 months, 6 months, and 12 months post-treatment to assess variceal eradication, rebleeding, and ulcer healing. Patients presenting with symptoms suggestive of rebleeding between scheduled visits underwent urgent endoscopy.

Observation indicators

(1) Comparison of treatment doses between the conventional endoscopy group and the EUS group: the doses of polidocanol and cyanoacrylate glue used, as well as the total number of treatments in each treatment technique. (2) Comparison of rebleeding rates between the conventional endoscopy group and the EUS group: The number of bleeding cases at 3 months, 6 months, and 1 year postoperatively, was compared, and the rebleeding rate was calculated. (3) Comparison of ulcer formation rates between the conventional endoscopy group and the EUS group: The number of glue expulsion ulcers formed at 3 months postoperatively, was calculated, and the ulcer formation rate was calculated (Fig. 5).

Fig. 5.

On review 3 months later, the original varicose vein mass disappeared, and the gastric fundus was seen to be drained of glue and scar-like changes

Endpoint definitions

Rebleeding was defined as any episode of hematemesis, melena, or hematochezia accompanied by a decrease in hemoglobin level of ≥ 2 g/dL within 24 h, or requiring hospital admission for gastrointestinal bleeding. All suspected rebleeding episodes were confirmed by urgent endoscopy, which documented active bleeding from gastroesophageal varices (Forrest Ia/Ib) or the presence of fresh thrombus overlying varices. Bleeding originating from other sources, such as portal hypertensive gastropathy, gastric ulcers, or erosive gastritis, was excluded from the rebleeding endpoint.

Ulcer formation referred specifically to mucosal defects occurring at the site of previous cyanoacrylate injection (glue expulsion ulcers). Ulcers were diagnosed endoscopically according to standard morphological criteria (crater with exudate or erythema) during scheduled follow-up examinations, primarily at 3 months post-procedure.

Statistical analysis

Statistical analysis was conducted via SPSS 27.0 software. For continuous data, normality tests were performed. If the data followed a normal distribution, statistical descriptions are given as the means ± standard deviations (x̅ ± s), and comparisons between two groups were conducted via t test. For categorical data (e.g., bleeding rate, and ulcer formation rate), the χ² test was used for statistical description, expressed as counts (%), with P < 0.05 indicating a statistically significant difference. Throughout the manuscript, only P values < 0.05 are described as statistically significant; P values ≥ 0.05 are referred to as trends or numerical differences, without implying statistical significance.

To adjust for baseline imbalances due to the non-randomized nature of the study, Propensity Score Matching (PSM) was conducted to minimize selection bias. Variables included in the matching model were age, gender, Child-Pugh classification, Sarin classification, and variceal grading. A 1:1 nearest-neighbor matching algorithm was used with a caliper width of 0.2 standard deviations of the logit of the propensity score. Standardized mean differences (SMDs) were used to assess balance, with SMD < 0.1 indicating good balance.

Results

Using either conventional endoscopy or EUS endoscopy, both groups of patients achieved effective occlusion of the target varices after treatment, and the surgical success rate was 100% in both groups (Table 1). No severe complications, such as organ embolism or death, were observed in either group. No thromboembolic events were observed during hospitalization or follow-up. Embolism was assessed clinically and, when indicated, by imaging; no asymptomatic embolisms were detected during follow-up. However, postoperative adverse events, such as fever, abdominal pain, melena, or hematochezia, occurred more frequently in the conventional endoscopy group than in the EUS group. In the conventional endoscopy group, there were 9 cases of melena or hematochezia (17.31%) after surgery, whereas there was only 1 case (1.81%) in the EUS group, and the difference was statistically significant (p = 0.016). The endoscopic treatment costs for the two groups were 15197.72 (12073.29,21446.58) RMB and 14048.86 (10924.43,17948.86) RMB, respectively, with no statistically significant difference (p > 0.05) (Table 1).

Table 1.

Outcome measures

	Conventional Endoscopy Group (N = 52)	EUS Group (N = 55)	P value
Surgical Success Rate [n (%)]	52(100.00%)	55(100.00%)	1.00
Adverse Events [n (%)]
Fever	2(3.85%)	1(1.81%)	0.961
Abdominal Pain	3(5.77%)	2(3.63%)	0.949
Melena or Hematochezia	9(17.31%)	1(1.81%)	0.016*
Organ Embolism	0	0	0
Death	0	0	0
Total Treatment Cost (RMB)	15197.72(12073.29, 21446.58)	14048.86(10924.43, 17948.86)	0.201

* p < 0.05 ** p < 0.01

Comparison of baseline data between the conventional endoscopy group and the EUS group

The baseline data of patients in both groups (age, sex, etiology, Child‒Pugh classification, Sarin classification, and variceal classification) were compared. There were no statistically significant differences in age, sex, etiology, Child‒Pugh classification, Sarin classification, or esophageal or gastric variceal classification across the seven items between the different groups (P > 0.05), indicating no statistical significance (Table 2).

Table 2.

Baseline characteristics of the conventional endoscopy and EUS groups

Indicator/Group		conventional endoscopy group N = 52	EUS Group N = 55	P value
Age		62.06 ± 10.72	59.38 ± 11.05	0.207
Gender	Male	25(48.10%)	32(58.20%)	0.295
	Female	27(51.90%)	23(41.80%)
Etiology	Viral hepatitis cirrhosis	32(61.50%)	42(76.40%)
	Alcoholic hepatitis cirrhosis	3(5.80%)	3(5.50%)	0.214
	Other types of hepatitis cirrhosis	17(32.70%)	10(18.20%)
Child‒Pugh Classification	A	11(21.20%)	9(16.40%)
	B	33(67.30%)	41(74.50%)	0.433
	C	8(15.40%)	5(9.10%)
Sarin Classification	GOV-1 type	17(32.70%)	11(20.00%)	0.135
	GOV-2 type	35(67.30%)	44(80.00%)
Varices Grading	Mild	0(0.00%)	0(0.00%)
	Moderate	22(42.30%)	25(45.50%)	0.743
	Severe	30(57.70%)	30(54.50%)

Comparison of hemoglobin and liver function levels within one month after surgery between the conventional endoscopy group and the EUS group

There were no significant differences in aspartate aminotransferase (AST), alanine aminotransferase (ALT), total bilirubin, or prothrombin time between the two groups (P > 0.05). After conventional endoscopy, hemoglobin (P < 0.01), albumin (P = 0.017), aspartate transaminase (AST) (P < 0.01), and alanine transaminase (ALT) (P < 0.01) levels significantly changed from pretreatment levels. After EUS treatment, hemoglobin (P < 0.01) and albumin levels (P < 0.01) also significantly differed from the pretreatment levels. As shown in the charts, there were no significant changes in total bilirubin or coagulation parameters before and after treatment in either the conventional endoscopy group or the EUS group (Table 3).

Table 3.

Baseline and postoperative liver function and laboratory parameters in the conventional endoscopy and EUS groups

Monitoring Indicator	Group	Conventional Endoscopy Group N = 52	EUS Group N = 55	P value
Hemoglobin(g/L)	Pre-op	83.71 ± 15.75	82.02 ± 13.33	0.549
	Post-op	91.87 ± 12.82	90.89 ± 12.39	0.690
	P value	0.000**	0.000**
Albumin(g/L)	Pre-op	31.72 ± 5.35	30.98 ± 5.49	0.499
	Post-op	32.93 ± 4.88	34.97 ± 5.00	0.306
	P value	0.017*	0.000**
Aspartate Aminotransferase (IU/L)	Pre-op	41.50(29.25,60.25)	34.00(27.00,53.00)	0.151
	Post-op	37.00(27.00,52.00)	34.00(25.00,45.00)	0.519
	P value	0.000**	0.082
Alanine Aminotransferase (IU/L)	Pre-op	29.00(21.25,42.50)	26.00(22.00,41.00)	0.556
	Post-op	22.50(14.25,37.00)	26.00(19.00,32.00)	0.337
	P value	0.000**	0.055
Total Bilirubin	Pre-op	21.80(14.20,29.28)	22.40(17.60,30.70)	0.271
(umol/L)	Post-op	20.35(15.05,30.08)	22.40(16.90,32.50)	0.491
	P value	0.405	0.406
Prothrombin Time	Pre-op	14.65(13.93,16.05)	14.60(13.60,16.30)	0.830
(s)	Post-op	14.00(13.03,15.00)	14.20(13.40,15.60)	0.305
	P value	0.092	0.053

Data are presented as mean ± SD or median (IQR). AST aspartate aminotransferase, ALT alanine aminotransferase.* p < 0.05 ** p < 0.01

Comparison of surgical treatment between the conventional endoscopy group and the EUS group

There were no significant differences in the total number of treatment sessions between the two treatment methods (P > 0.05), indicating no statistical significance. However, the amounts of polidocanol and cyanoacrylate glue used in the conventional endoscopy group were significantly greater than those used in the EUS group (P < 0.05) (Table 4).

Table 4.

Surgical treatment comparison between endoscopy group and EUS group

	Conventional Endoscopy Group (N = 52)	EUS Group (N = 55)	T value	P value
Total Volume of Polidocanol (mL)	44.54 ± 23.92	10.96 ± 6.37	9.798	0.000**
Total Volume of cyanoacrylate (mL)	4.15 ± 1.90	2.78 ± 1.88	3.747	0.001**
Number of Treatments	1.79 ± 0.87	1.53 ± 0.77	1.649	0.102

* p < 0.05 ** p < 0.01

Comparison of the rebleeding rate and ulcer formation rate between the conventional endoscopy group and the EUS group

Both groups achieved effective hemostasis after treatment and were followed up regularly. Although numerically higher in the EUS group at 3 months, this difference was not statistically significant (P = 0.281). Rebleeding rates at 3 months and 1 year postoperatively were lower in the EUS group than in the conventional endoscopy group. Furthermore, the 6‑month rebleeding rate and the 3‑month ulcer formation rate were significantly lower in the EUS group (both P < 0.05; Table 5). The most pronounced benefit of EUS‑SVD emerged at 6 months, suggesting a time‑dependent treatment effect.

Table 5.

Comparison of postoperative rebleeding rate after treatment and ulcer formation rate between conventional endoscopy group and EUS group

Group	N	Rebleeding at 3 Months	Rebleeding at 6 Months	Rebleeding at 1 Year	Ulcer Formation at 3 Months
Conventional Endoscopy Group [n (%)]	52	13(25.00%)	16(30.80%)	13(25.00%)	15(28.85%)
EUS Group [n (%)]	55	19(34.50%)	3(5.50%)	6(10.90%)	7(12.73%)
χ² Value		1.162	10.060	3.634	4.252
P Value		0.281	0.002**	0.057	0.039*

* p < 0.05 ** p < 0.01

Kaplan‒Meier analysis was conducted to compare the cumulative incidence of rebleeding between the two groups (Fig. 6). The cumulative incidence of rebleeds was significantly lower in the EUS group (HR 0.38, 95%CI 0.17–0.86, P = 0.021) than in the conventional endoscopy group. The mean time to rebleeding in the conventional endoscopy group was 394 days (95% CI 197–591 days); in the EUS group, it was 464 days (95% CI 411–516 days).

Fig. 6.

Kaplan‒Meier curves for the cumulative incidence of rebleeding in the two groups. The log-rank test suggested a significant difference in the cumulative incidence of reintervention between the groups (log-Rank P = 0.021)

Rebleeding rate and ulcer formation rate in GOV-1 and GOV-2 after different treatment methods

Patients were divided into two groups according to the Sarin classification: the GOV-1 group and the GOV-2 group. This classification was proposed by Indian scholar Sarin and describes gastric varices (GV) and esophageal varices (EV) on basis of their relationship within the stomach and the location of the GV in the stomach. It can be divided into gastroesophageal varices (GOVs) and isolated gastric varices (IGVs). The GOV types are further divided into GOV1 and GOV2 types. GOV1 refers to a GV located in the lesser curvature of the stomach, combined with the presence of an EV. GOV2 refers to a GV located in the greater curvature of the stomach and is also combined with an EV. In this study, patients were divided into two groups: GOV-1 group and the GOV-2 group. Both groups achieved effective hemostasis after treatment and were followed up regularly.

Patients in the GOV-1 group were categorized to receive two different treatments and were divided into the conventional endoscopy group and the EUS group. No statistically significant differences were observed in rebleeding rates at 3, 6, or 12 months or in ulcer formation rate in the GOV-1 subgroup (all P > 0.05), although numerically lower rates were noted in the EUS group (Table 6).

Table 6.

 Comparison of bleeding rate and ulcer formation rate between conventional endoscopy group and EUS group for GOV-1 type

Group	N	Rebleeding at 3 Months	Rebleeding at 6 Months	Rebleeding at 1 Year	Ulcer Formation at 3 Months
Conventional Endoscopy Group [n (%)]	17	3(17.65%)	5(29.41%)	3(17.65%)	5(29.41%)
EUS Group [n (%)]	11	4(36.36%)	0(0.00%)	1(9.09%)	1(9.09%)
χ² Value
P Value		0.381	0.125	1.000	0.355

In the GOV-2 group, patients were categorized to two different treatment groups, forming the conventional endoscopy group and the EUS group. The comparison of the rebleeding rates at 6 months and 1 year postoperatively between the two groups revealed statistically significant differences (P < 0.05), with the rebleeding rates at 6 months and 1 year postoperatively in the EUS group being significantly lower than those in the conventional endoscopy group. The ulcer formation rate at 3 months postoperatively was lower in the EUS group than in the conventional endoscopy group, but the difference was not statistically significant (P > 0.05). On the other hand, the rebleeding rate at 3 months postoperatively in the conventional endoscopy group was lower than that in the EUS group, but the difference in the rebleeding rates at 3 months postoperatively between the two groups was also not statistically significant (P > 0.05) (Table 7).

Table 7.

 Comparison of bleeding rate and ulcer formation rate between conventional endoscopy group and EUS group for GOV-2 type

Group	N	Rebleeding at 3 Months	Rebleeding at 6 Months	Rebleeding at 1 Year	Ulcer Formation at 3 Months
Conventional Endoscopy Group [n (%)]	35	10(28.57%)	11(31.43%)	11(31.43%)	10(28.57%)
EUS Group [n (%)]	44	15(34.09%)	3(6.82%)	5(11.36%)	6(13.64%)
χ² Value		0.275	8.097	4.859	2.692
P Value		0.600	0.004**	0.027*	0.101

* p < 0.05 ** p < 0.01

Comparison of rebleeding rate and ulcer formation rate after different treatment methods in patients with moderate and severe varices

Patients were divided into moderate and severe varices groups on the basis of the severity of the varices. Both groups achieved effective hemostasis after treatment and were followed up regularly. In the moderate varices group, patients were categorized to two different treatment group, forming the conventional endoscopy group and the EUS group. In patients with moderate varices, no statistically significant differences were found between groups (all P > 0.05), although the EUS group showed lower rates of rebleeding and ulcer formation (Table 8).

Table 8.

Comparison of bleeding rate and ulcer formation rate between standard endoscopy group and EUS group for moderate varices

Group	N	Rebleeding at 3 Months	Rebleeding at 6 Months	Rebleeding at 1 Year	Ulcer Formation at 3 Months
Conventional Endoscopy Group [n (%)]	22	6(27.27%)	7(31.82%)	3(13.64%)	7(31.82%)
EUS Group [n (%)]	25	9(36.00%)	2(8.00%)	6(24.00%)	6(24.00%)
χ² Value		0.410	2.888	0.280	0.357
P Value		0.522	0.089	0.596	0.550

In the severe varices group, patients were categorized to two different treatment groups, forming the conventional endoscopy group and the EUS group. The rebleeding rates at 6 months and 1 year postoperatively, as well as the ulcer formation rate at 3 months postoperatively, were significantly lower in the EUS group than in the conventional endoscopy group. The comparison of rebleeding rates at 6 months and 1 year postoperatively and the ulcer formation rate at 3 months postoperatively between the two groups revealed statistically significant differences (P < 0.05). However, the difference in rebleeding rates at 3 months postoperatively between the two groups was not statistically significant (P > 0.05), with the EUS group having a higher rebleeding rate at 3 months postoperatively than the conventional endoscopy group did (Table 9).

Table 9.

 Comparison of bleeding rate and ulcer formation rate between conventional endoscopy group and EUS group for severe varices

Group	N	Rebleeding at 3 Months	Rebleeding at 6 Months	Rebleeding at 1 Year	Ulcer Formation at 3 Months
Conventional Endoscopy Group [n (%)]	30	7(23.33%)	9(30.00%)	11(36.67%)	8(26.67%)
EUS Group [n (%)]	30	10(33.33%)	1(3.33%)	0(0.00)	1(3.33%)
χ² Value		0.739	7.680	13.469	6.405
P Value		0.390	0.006**	0.000**	0.011*

* p < 0.05 ** p < 0.01

Univariate and multivariate logistic regression analysis of rebleeding from varices at 6 months postoperatively

To further explore factors associated with early rebleeding, univariate and multivariable logistic regression analyses were performed (Table 10). Given the limited number of rebleeding events (n = 19) relative to the number of candidate predictors, we prioritized model parsimony to reduce the risk of overfitting. Consequently, the multivariable model was restricted to three clinically relevant variables: age, Child–Pugh classification, and endoscopic treatment modality.

Table 10.

Univariate and multivariate logistic regression analysis associated with postoperative rebleeding within six months

Variable	univariate		multivariate
	OR(95%CI)	P value	OR(95%CI)	P value
Age	1.040(0.991–1.091)	0.112	1.116(1.034–1.203)	0.005**
Child-Pugh Classification	0.690(0.279–1.705)	0.422	0.511(0.159–1.641)	0.260
Endoscopic Treatment Method (Ref: Conventional)	0.130(0.035–0.478)	0.002	0.05(0.009–0.263)	<0.001**

* p < 0.05 ** p < 0.01

In the multivariable analysis, EUS-SVD remained independently associated with a significantly lower risk of 6-month rebleeding (adjusted OR 0.05, 95%CI 0.009–0.263, P < 0.001). Older age was also independently associated with an increased risk of rebleeding (adjusted OR 1.12 per year, 95%CI 1.03–1.20, P = 0.005). Child–Pugh classification did not retain statistical significance in this model.

These findings should be interpreted with caution. The small number of events inherently limits the statistical power and stability of multivariable estimates. Therefore, the observed associations are best regarded as hypothesis-generating rather than definitive evidence of causality. Larger prospective studies are required to confirm these relationships.

Discussions

Gastroesophageal varices (GOVs) are a common clinical manifestation in patients with portal hypertension and are a frequent complication of cirrhosis. The risk of variceal rupture is closely related to the severity of cirrhosis [10]. This condition is also a common cause of upper gastrointestinal bleeding, second only to duodenal and gastric ulcers. Prolonged portal hypertension weakens the vessel walls of varices, making them more prone to rupture because of factors such as gastric acid erosion, mechanical friction from food, or a sudden increase in intra-abdominal pressure [11]. Once rupture occurs, the bleeding is often rapid and recurrent, even if successful hemostasis is achieved during emergency care. The rebleeding rate can still reach 70%-80%, and the mortality rate from rebleeding is as high as 25%-50% [12, 13]. Therefore, gastrointestinal endoscopy has become the first-line treatment for emergency hemostasis and rebleeding prevention in cases of GOV. Clinically, endoscopic sclerotherapy combined with tissue adhesive injection is commonly used. The tissue adhesive reacts with blood to solidify quickly, forming a clot that blocks the ruptured vessel, thereby achieving hemostasis and preventing rebleeding. However, since gastric varices are located in the submucosa, distinguishing between varices and gastric folds via conventional endoscopy is difficult. If sclerotherapy and adhesive injections are performed under conventional endoscopy, there is a risk of injection adjacent to the varices, increasing the risk of complications. Therefore, while conventional endoscopic sclerotherapy and adhesive injection are currently among the first-choice treatments for GOV, precise injection into the varices is necessary to ensure complete occlusion, thereby achieving high hemostasis rates while reducing rebleeding and complications.

Within the limitations of this single-center retrospective study, the advantages of endoscopic ultrasound (EUS)-guided treatment are more pronounced than those of conventional endoscopy, including precise localization of the varices, ultrasound visualization throughout the injection process, and the ability to confirm variceal occlusion through Doppler imaging. It should be noted that EUS-guided intervention requires advanced endoscopic skills and familiarity with Doppler ultrasound interpretation, which may limit its immediate applicability in centers without specialized expertise. According to Boustiere et al., EUS increases the detection rate of varices sixfold [14]. Iwase et al. also reported that residual blood flow after endoscopic treatment was closely related to the risk of rebleeding [15]. EUS can also measure variceal diameter to control the amount of tissue adhesive injected.

Meanwhile, recent high-level evidence has consistently highlighted the evolving role of EUS-guided interventions. Meta-analyses by Baig et al. and Mohan et al. [16, 17]have shown that EUS-guided coil plus cyanoacrylate improves variceal obliteration and reported trends toward lower rebleeding compared with conventional endoscopic glue injection, albeit with substantial heterogeneity. Supporting this trajectory, the network meta-analysis by Giri et al. suggested that BRTO and EUS-guided therapies may rank superior to direct glue injection for specific outcomes, although the overall certainty of evidence remains limited [18]. Moreover, while the Cochrane review by Jackson et al. confirmed the superiority of cyanoacrylate over band ligation for rebleeding reduction, it did not capture the emerging EUS-based techniques, underscoring the rapid evolution of this field [19]. Contemporary technical reviews and observational studies from 2020 to 2024 have further refined these concepts, describing various EUS-guided approaches, coil-glue combinations, and their favorable safety profiles.

Therefore, EUS-guided sclerotherapy and adhesive injection may be safer and more effective treatment options for patients with GOV. In our study, we used EUS to scan and measure the diameter and connections of varices to determine whether to use coil embolization. We then performed the “sugar‒glue‒sugar” sandwich method under EUS guidance. The advantages of this method include the following: (1) Precise localization of the varices, including their diameter and connections. (2) Real-time EUS scanning to ensure that the needle enters the varices. (3)”Ultrasound visualization” of the entire injection process, monitoring changes in the vessel lumen and the disappearance of blood flow. (4) Reduced tissue adhesive dosage, which lowers the risk of ectopic embolization and reduces treatment costs.

To ensure the accuracy of the data, we did not administer postoperative β-blockers therapy for primary prevention. The results of this study show that both the conventional endoscopy group and the EUS group achieved effective hemostasis after treatment and regular follow-up. Compared with the conventional endoscopy group, the EUS group received less sclerosing agent, tissue adhesive, and required fewer treatment sessions. Additionally, EUS was associated with a reduction in rebleeding at 3 months and 1 year, although these differences did not reach statistical significance. The most robust benefit was observed at 6 months, consistent with more complete varix obliteration. These findings suggest that compared with conventional endoscopy, that EUS-guided sclerotherapy and adhesive injection require fewer materials during surgery and result in lower rebleeding and ulcer formation rates. EUS provides precise ultrasound scanning of the gastric lumen, revealing the detailed structure of the gastric wall, and color Doppler flow imaging allows for clear and vivid visualization of the tortuous veins. Under EUS guidance, variceal vessels can be accurately localized, reducing the likelihood of off-target injections seen in conventional endoscopy. Furthermore, EUS allows for precise determination of whether the varices have been completely occluded, reducing the risk of rebleeding. EUS also offers the ability to visualize concealed or hard-to-detect bleeding points, providing a solution to the limitations of traditional diagnostic and therapeutic methods. This strengthens the technological support for emergency clinical interventions. In this study, we compared treatment outcomes between the conventional endoscopy group and the EUS group for different types of GOVs and varying degrees of varices. The results revealed that for patients with GOV1, GOV2, or moderate to severe varices, the rebleeding rates at 6 and 12 months and the ulcer formation rates at 3 months were lower in the EUS group than in the conventional endoscopy group. This demonstrates that EUS results in a higher variceal occlusion rate than conventional endoscopy does, resulting in a lower rate of late rebleeding [20, 21].

Owing to the limited number of cases in this study, there may be a lack of statistical power, meaning that potential differences in adverse events between treatment groups may not have been fully revealed. Whether EUS should be widely applied in precise devascularization surgery requires further collection of large amounts of data for comparative studies. Another limitation is the small number of rebleeding events, which restricted our ability to perform robust multivariate logistic regression analysis. The odds ratios for some variables, particularly etiology and treatment modality, were notably small, potentially reflecting overfitting and instability rather than strong biological effects. Future studies with larger sample sizes are required to validate these associations.

Limitations of the study

First, as a single-center retrospective cohort, selection bias and unmeasured confounding cannot be excluded despite PSM.

Second, all subgroup analyses were exploratory, and multiple comparisons were not adjusted; therefore, these findings should be considered hypothesis-generating rather than confirmatory.

Third, the limited number of rebleeding events restricted our ability to perform robust multivariable logistic regression analysis. To minimize overfitting, we adopted a parsimonious model including only core clinical covariates; nevertheless, these regression results remain exploratory.

Fourth, external validity may be limited to centers with similar EUS expertise.

Conclusions

Previous studies have shown that endoscopic ultrasound selective variceal devascularization for gastroesophageal varices has a good therapeutic effect on variceal rupture and bleeding. The aim of this study was to understand the application value of precise endoscopic ultrasound-guided disconnection in the treatment of esophageal and gastric varices caused by portal hypertension and to provide a new approach for clinical diagnosis and treatment. We identified that the clinical outcomes are significant and that the postoperative complications are mild, with most being effectively alleviated through symptomatic treatment.

In conclusion, within the limitations of this single-center retrospective study, EUS-SVD appears to be associated with reduced rebleeding and ulcer formation rates compared with conventional endoscopy. However, given the non-randomized design and potential for operator-dependent variability, these findings should be interpreted as preliminary. Further large-scale, prospective, randomized controlled trials are required to validate the efficacy and safety of EUS-SVD before it can be broadly recommended for routine clinical practice.

Abbreviations

ALT

Alanine aminotransferase

AST

Aspartate aminotransferase

CI

Confidence interval

EUS-SVD

Endoscopic ultrasound selective variceal devascularization

GOV

Gastroesophageal varices

GOV1

Gastroesophageal varices type 1 (Sarin classification)

GOV2

Gastroesophageal varices type 2 (Sarin classification)

HR

Hazard ratio

IGV

Isolated gastric varices

OR

Odds ratio

PH

Portal hypertension

PSM

Propensity score matching

RCT

Randomized controlled trial

SMD

Standardized mean difference

Authors’ contributions

Fan Huang: Writing-original draft, Visualization, Software, Methodology, Data curation. Chengqi Guan: Writing-review & editing, Project administration, Funding acquisition. Hongyuan Shen: Methodology, Investigation, Formal analysis. YaChen Wang: Software, Methodology. Wei Huang: Supervision, Formal analysis. Cuihua Lu: Project administration, Conceptualization. Zhenbiao Mao: Investigation, Data curation. Jianfeng Zhang: Writing-review & editing, Validation, Resources, Funding acquisition, Conceptualization.

Funding

The present study was supported by the Nantong 14th Five Year Plan for Young Medical Talents Project (NTQNRC146) and the Postdoctoral Research Project of Nantong University Affiliated Hospital (BSH202214).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

The Ethics Committee of Nantong University Affiliated Hospital approved this retrospective study (2020-L093) and waived the requirement for written informed consent due to the retrospective nature of the study.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.