KSY and KSJ contributed equally to this study.
Transcatheter arterial embolization (TAE) is useful for management of uncontrolled upper gastrointestinal (UGI) bleeding. We investigated clinical outcomes of TAE for non-variceal bleeding from benign UGI diseases uncontrolled with endoscopic intervention.
This retrospective study performed between 2017 and 2021 across four South Korean hospitals. Ninety-two patients (72 men, 20 women) who underwent angiography were included after the failure of endoscopic intervention for benign UGI disease- induced acute non-variceal bleeding. We investigated the factors associated with endoscopic hemostasis failure, the technical success rate of TAE, and post-TAE 30-day rebleeding and mortality rates.
The stomach (52/92, 56.5%) and duodenum (40/92, 43.5%) were the most common sites of bleeding. Failure of endoscopic procedures was attributable to peptic ulcer disease (81/92, 88.0%), followed by pseudo-aneurysm (5/92, 5.4%), and angiodysplasia (2/92, 2.2%). Massive bleeding that interfered with optimal visualization of the endoscopic field was the most common indication for TAE both in the stomach (22/52, 42.3%) and duodenum (14/40, 35.0%). Targeted TAE, empirical TAE, and exclusive arteriography were performed in 77 (83.7%), nine (9.8%), and six patients (6.5%), respectively. The technical success rate, the post-TAE 30-day rebleeding rate, and the overall mortality rate were 100%, 22.1%, and 5.8%, respectively. On multivariate analysis, coagulopathy (OR, 5.66; 95% CI, 1.71~18.74;
TAE may be useful for acute non-variceal UGI bleeding. Targeted embolization and correction of coagulopathy can improve clinical outcomes.
Upper gastrointestinal (UGI) bleeding includes both variceal and non-variceal bleeding. Nonvariceal bleeding includes bleeding associated with peptic ulcers, esophageal-gastric junction mucosal lacerations, and vascular dysplasia [
The purpose of this multicenter study was to identify the clinical situations of angiography and vascular embolization performed in Korea. We also analyzed the clinical outcomes, such as the technical success, rebleeding, and 30-day mortality rates. The factors affecting the occurrence of rebleeding within 30 days of vascular embolization were also investigated.
This study retrospectively analyzed the medical records of patients who underwent angiography and embolization for acute nonvariceal UGI bleeding, after the failure of UGI endoscopic hemostasis, at Seoul St. Mary's Hospital, Keimyung University Dongsan Hospital, Yangsan Pusan National University Hospital, and Myongji Hospital between January 2017 and December 2021. Eligible patients were identified by the endoscopy and radiology intervention team of the departments of Gastroenterology and Radiology. A total of 92 patients with acute nonvariceal UGI bleeding were analyzed. Patients were excluded if they had bleeding from the esophagus or gastricvarices. Patients with cancer bleeding were also excluded. This study was approved by the Institutional Bioethics Committee of Myongji Hospital (No. 2022-03-027).
UGI endoscopy was first performed in the presence of bleeding symptoms, including vomiting, black stool, and bloody stool, and decreased hemoglobin levels. Abdominal CT was performed if the bleeding continued due to unsuccessful hemostasis or if the source of the bleeding could not be clearly identified due to severe bleeding.
Angiography was performed, through the femoral artery, to identify bleeding sites in the celiac, superior mesenteric, and splenic arteries. Targeted embolization was performed when angiography confirmed outflow of contrast agent or if there were indirect signs of bleeding, including pseudoaneurysm formation, vascular irregularities, vascular blockage, increased numbers of blood vessels in an area of neovascularization, and dilated arterioles. Empiric embolization was performed in the absence of angiographically confirmed bleeding lesions if the UGI endoscopy and abdominal CT findings were indicative of bleeding.
Each embolic agent, a fine coil (Tornado or MicroNester; Cook Medical, Bloomington, IN, USA), Gelatin Sponge (Spongostan; Ethicon, Somerville, NJ, USA and EG-GEL; Engain, Seongnam, Korea), N-butyl-cyanoacrylate glue (Glubran; GEM SRL, Viareggio, Italy and Histoacryl®; B. Braun, Melsungen, Germany), Pamiray300 (Dongkook Pharm., Seoul, Korea and Therapex; Montreal, Canada), or polyvinyl alcohol particles (Contour; Boston Scientific, Watertown, MA, USA and Bearing nsPVA; Merit Medical System, South Jordan, UT, USA) was used alone or in combination with other agents. Angiography and vascular embolization were performed by 19 radiologists with 3~22 years of experience on interventional radiology procedures.
The enrolled patients’ medical records, upper intestinal tract endoscopy results, blood tests, CT findings, vital signs, angiography and vascular embolization images, and results were checked. The causes of UGI bleeding, endoscopic hemostasis failure, and reasons for angiography were investigated. The success of the vascular embolization (hemostasis) was confirmed.
UGI bleeding was defined as bleeding that occurred proximal to the ligament of Treitz. Technical success was defined as occlusion of the blood vessels associated with the bleeding site, cessation of contract agent leakage, or absence of pseudoaneurysm formation after embolization. Rebleeding was defined as the recurrence of clinical bleeding symptoms, such as vomiting, black stools, or bloody stools, within 30 days of the embolization, accompanied by a ≥2.0 g/dL decrease in hemoglobin levels. Among the potential vascular embolization-related complications, gastric ischemic damage was confirmed by endoscopic evidence of ischemic damage centered on the mucosal layer of the UGI tract; splenic infarction was confirmed by decreased contrast enhancement or loss of abdominal CT findings. Coagulopathy was defined as a prothrombin international normalized ratio ≥1.5, partial thromboplastin time >45 seconds, or a platelet count <80,000/mL [
Statistical analyses were performed using SPSS for Windows software (version 21.0; SPSS, Chicago, IL, USA). Continuous variables were analyzed using student's t-test or the Kruskal-Wallis test to determine mean±standard deviations and medians (minimum-maximum). Categorical variables, expressed as frequencies (%), were analyzed using the chi-square test or Fisher's exact test. To identify the independent factors associated with rebleeding within 30 days of vascular embolization, variables with a P-value <0.10 in a univariate analysis were included in a multivariate logistic regression model. If the P-value was less than 0.05, it was judged to be statistically significant.
The endoscopically determined cause of the bleeding that led to angiography was peptic ulcers in 81 patients (88.0%) and pseudoaneurysms in five (5.4%). Less frequent causes included vascular dysplasia, dieulafoy lesion, duodenal diverticulum, bleeding after endoscopic submucosal dissection of the stomach, and iatrogenic damage during endoscopy (
Of the 92 patients who underwent angiography, 77 underwent selective embolization when hemorrhagic lesions were identified. In nine of the 15 patients with bleeding lesions that could not be identified using angiography, empirical embolization was performed at the suspected bleeding site, according to clinical judgment. In six patients, angiographic examinations were performed without embolization (
Angiography was performed within 12 hours after endoscopic hemostasis failure in 43 patients (46.7%), within 12~24 hours in 22 patients (23.9%), and after 24 hours in 27 patients (29.3%) (
Except for six patients who underwent angiography, without vascular embolization, the technical success rate of the vascular embolizations in the remaining 86 patients was 100%. A total of 19 patients (22.1%) demonstrated rebleeding within 30 days of the vascular embolization and five (5.8%) died within the initial 30-day period (
The incidence of complications associated with vascular embolization was 3.5% (3/86). All complications occurred after embolization of gastric bleeding lesions (
On multivariate logistic regression analysis, risk factors for rebleeding within 30 days after vascular embolization were coagulation disorders (OR, 5.66; 95% CI, 1.71~18.74;
This study is the first multicenter study in Korea that analyzed the use of angiography and embolization after UGI endoscopic hemostasis for acute nonvariceal gastrointestinal bleeding. Clinical performance of vascular embolization was also evaluated. In cases of acute nonvariceal gastrointestinal tract bleeding, the causes of endoscopic hemostasis failure were analyzed after classifying the cases into gastric and duodenal bleeding. The most common reason for endoscopic hemostasis failure was associated with difficult visualization of the bleeding source because of excessive bleeding from the gastric or duodenal ulcer disease. In this study, the peptic ulcer bleeding foci that led to vascular embolization after endoscopic hemostasis failure were analyzed using the Forrest classification. Forrest IIa was most common in gastric ulcers and Forrest Ib was most common in duodenal ulcers. Forrest IIb ulcers were associated with disagreements regarding the need for endoscopic hemostasis (four cases each of gastric and duodenal ulcers). None of the cases corresponded to Forrest IIc and III ulcers that did not require endoscopic hemostasis. Rebleeding rates are approximately 55% for Forrest Ia-Ib ulcers, 43% for Forrest IIa, 25% for Forrest IIb, <10% for Forrest IIc, and <5% for Forrest III ulcers [
In previous studies, the rate of detection of hemorrhagic lesions during angiography is in the range of 45.9~86.4% [
Worldwide, radiological vascular intervention is becoming a common alternative treatment for refractory nonvariceal UGI bleeding [
A multivariate analysis of the occurrence of rebleeding within 30 days of vascular embolization confirmed that coagulation disorders and empiric embolization were significant contributing factors. Previous studies also indicated that coagulation disorders are significantly associated with an increased risk of rebleeding after vascular embolization [
Complications associated with vascular embolization occur in 0~18.9% of cases. Further, because the UGI tract has collateral blood circulation, the frequency of complications is lower than following embolization in the central or lower gastrointestinal tract [
This study has some limitations. First, this study was based on a retrospective analysis of medical records; therefore, there is a possibility of selection bias. Further, patients did not necessarily receive similar examinations and treatment owing to the multicenter nature of the study. Second, because the experience and skills of the endoscopists and interventional radiologists varied, follow-up studies are needed to generalize the findings. Third, among patients with acute nonvariceal UGI bleeding, patients with malignant upper intestinal bleeding (e.g., gastric cancer) were excluded. This was because, unlike benign nonvariceal gastrointestinal bleeding, patients with malignant UGI bleeding may also have had direct invasion of blood vessels, severe mucositis, or tissue necrosis due to chemotherapy or radiation. Hence, there is a large difference in hemostatic treatment methods and the risk of rebleeding [
In summary, the causes of acute, nonvariceal UGI bleeding in patients who underwent angiography and vascular embolization following endoscopic hemostasis failure in Korea, were peptic ulcers and pseudoaneurysms. The technical success rate for vascular embolization in these patients was excellent, but rebleeding within 30 days of the procedure occurred in approximately 22% of cases. Since coagulation disorders and empirical embolization were significant factors associated with the occurrence of rebleeding, confirming and correcting coagulation disorders are necessary. Moreover, identifying the etiology of the bleeding (e.g., using UGI endoscopy to find the bleeding lesions) would help to ensure more cases of selective embolization and maximize the number of longer-term successes.
The authors have no potential conflicts of interest to disclose.
Supplementary material 1. Korean translation of the article is available from
Characteristics of endoscopic hemostasis based on the bleeding site.
(A) Results of angiography after failed endoscopic hemostasis performed for nonvariceal upper gastrointestinal bleeding. (B) Results of angiography after failed endoscopic hemostasis performed for nonvariceal gastric bleeding. (C) Results of angiography after failed endoscopic hemostasis performed for nonvariceal duodenal bleeding. UGI, upper gastrointestinal.
Demographic Characteristics of Patients
Total (n=92) | Stomach (n=52) | Duodenum (n=40) | |||
---|---|---|---|---|---|
Sex, men | 72 (78.3) | 46 (88.5) | 26 (65.0) | 0.014 | |
Age (years) | 66.1±13.8 | 65.2±13.6 | 67.2±14.1 | 0.498 | |
Body mass index (kg/m2) | 22.2±4.0 | 22.1±3.9 | 22.4±4.2 | 0.698 | |
Types of chronic disease | |||||
≥2 chronic disease | 43 (46.7) | 24 (46.2) | 19 (47.5) | 1.000 | |
Hypertension | 38 (41.3) | 23 (44.2) | 15 (37.5) | 0.531 | |
Diabetes mellitus | 25 (27.2) | 13 (25) | 12 (30.0) | 0.641 | |
Ischemic heart disease | 4 (4.3) | 3 (5.8) | 1 (2.5) | 0.630 | |
Chronic liver disease | 19 (20.7) | 12 (23.1) | 7 (17.5) | 0.693 | |
End stage renal failure | 6 (6.5) | 1 (1.9) | 5 (12.5) | 0.082 | |
History of peptic ulcer disease | 18 (19.6) | 11 (21.2) | 7 (17.5) | 0.863 | |
Laboratory findings | |||||
Hemoglobin, g/dL | 6.6±1.4 | 6.6±1.1 | 6.5±1.8 | 0.711 | |
BUN, mg/dL | 55.7±24.8 | 54.2±20.2 | 57.6±30.0 | 0.533 | |
Use of antiplatelet agent | 18 (19.6) | 12 (23.1) | 6 (15.0) | 0.482 | |
Use of anticoagulant agent | 6 (6.5) | 3 (5.8) | 3 (7.5) | 1.000 | |
Use of non-steroidal anti-inflammatory drug | 5 (5.4) | 4 (7.7) | 1 (2.5) | 0.383 | |
Alcohol drinking | 36 (39.1) | 24 (46.2) | 12 (30.0) | 0.174 | |
Smoking | 30 (32.6) | 23 (44.2) | 7 (17.5) | 0.013 | |
Initial presentation of UGI bleeding | 0.324 | ||||
Melena | 31 (33.7) | 13 (25.0) | 18 (45.0) | ||
Hematochezia | 19 (20.7) | 11 (21.2) | 8 (20.0) | ||
Hematemesis | 22 (23.9) | 17 (32.7) | 5 (12.5) | ||
Melena + hematemesis | 14 (15.2) | 9 (17.3) | 5 (12.5) | ||
Hematochezia + hematemesis | 2 (2.2) | 0 (0.0) | 2 (5.0) | ||
Anemia | 4 (4.3) | 2 (3.8) | 2 (5.0) | ||
Etiology of bleeding | 0.561 | ||||
Peptic ulcer | 81 (88.0) | 45 (86.5) | 36 (90.0) | ||
Pseudoaneurysm | 5 (5.4) | 2 (3.8) | 3 (7.5) | ||
Angiodysplasia | 2 (2.2) | 2 (3.8) | 0 (0.0) | ||
Diverticular bleeding | 1 (1.1) | 0 (0.0) | 1 (2.5) | ||
Dieulafoy’s lesion | 1 (1.1) | 1 (1.9) | 0 (0.0) | ||
Post gastric ESD for gastric dysplasia | 1 (1.1) | 1 (1.9) | 0 (0.0) | ||
Iatrogenic injury | 1 (1.1) | 1 (1.9) | 0 (0.0) | ||
Coagulopathy | 24 (26.1) | 15 (28.8) | 9 (22.5) | 0.633 | |
Shock | 58 (63.0) | 38 (73.1) | 20 (50.0) | 0.040 | |
Bleeding score system | |||||
GBS score | 13.4±2.8 | 13.7±2.9 | 12.9±2.6 | 0.184 | |
AIMS-65 | 2.5±1.2 | 2.8±1.3 | 2.3±1.0 | 0.036 | |
Types of endoscopic hemostasis | 0.476 | ||||
Epinephrine injection only | 13 (14.1) | 9 (17.3) | 4 (10.0) | ||
Thermal therapy only | 11 (12.0) | 5 (9.6) | 6 (15.0) | ||
Hemoclipping only | 5 (5.4) | 1 (1.9) | 4 (10.0) | ||
Epinephrine injection + hemoclipping | 9 (9.8) | 5 (9.6) | 4 (10.0) | ||
Thermal therapy + hemoclipping | 14 (15.2) | 8 (15.4) | 6 (15.0) | ||
Epinephrine injection + hemostatic powder spray | 3 (3.3) | 3 (5.8) | 0 (0.0) | ||
Hemoclipping + hemostatic powder spray | 3 (3.3) | 2 (3.8) | 1 (2.5) | ||
None | 34 (37.0) | 19 (36.5) | 15 (37.5) | ||
Study before arteriography | 0.205 | ||||
EGD only | 27 (29.3) | 12 (23.1) | 15 (37.5) | ||
EGD + CT angiography | 2 (2.2) | 1 (1.9) | 1 (2.5) | ||
EGD + abdominal pelvis CT | 63 (68.5) | 39 (75.0) | 24 (60.0) | ||
CT findings | 0.253 | ||||
Not performed | 27 (29.3) | 12 (23.1) | 15 (37.5) | ||
Contrast extravasation | 26 (28.3) | 14 (26.9) | 12 (30.0) | ||
Indirect signs of hemorrhage | 16 (17.4) | 12 (23.1) | 4 (10.0) | ||
No active bleeding | 23 (25.0) | 14 (26.9) | 9 (22.5) |
Values are presented as mean±standard deviation or number (%).
BUN, blood urea nitrogen; UGI, upper gastrointestinal; ESD, endoscopic submucosal dissection; GBS, Glasgow-Blatchford bleeding score; EGD, esophagogastroduodenoscopy; CT, computed tomography.
Summary of Angiography and Transcatheter Arterial Embolization Procedures
Total (n=92) | Stomach (n=52) | Duodenum (n=40) | |||
---|---|---|---|---|---|
Timing of angiography | 0.956 | ||||
Emergent (<12 hour) | 43 (46.7) | 25 (48.1) | 18 (45.0) | ||
Early (12~24 hour) | 22 (23.9) | 12 (23.1) | 10 (25.0) | ||
Late (>24 hour) | 27 (29.3) | 15 (28.8) | 12 (30.0) | ||
Arteries embolized | 0.000 | ||||
Left gastric artery | 29 (31.5) | 29 (55.8) | 0 (0.0) | ||
Right gastric artery | 11 (12.0) | 11 (21.2) | 0 (0.0) | ||
Gastroepiploic artery | 2 (2.2) | 2 (3.8) | 0 (0.0) | ||
Gastroduodenal artery | 30 (32.6) | 0 (0.0) | 30 (75.0) | ||
Inferior pancreaticoduodenal artery | 5 (5.4) | 0 (0.0) | 5 (12.5) | ||
2 territories embolization | 6 (6.5) | 6 (11.5) | 0 (0.0) | ||
≥3 territories embolization | 3 (3.3) | 1 (1.9) | 2 (5.0) | ||
Angiography only | 6 (6.5) | 3 (5.8) | 3 (7.5) | ||
Types of embolic agents | 0.655 | ||||
Microcoils only | 4 (4.3) | 1 (1.9) | 3 (7.5) | ||
Gelatin sponge only | 45 (48.9) | 26 (50.0) | 19 (47.5) | ||
PVA particles only | 6 (6.5) | 5 (9.6) | 1 (2.5) | ||
NBCA only | 17 (18.5) | 11 (21.2) | 6 (15.0) | ||
Microcoils + gelatin sponge | 7 (7.6) | 3 (5.8) | 4 (10.0) | ||
Microcoils + PVA particles | 2 (2.2) | 1 (1.9) | 1 (2.5) | ||
Microcoils + NBCA | 3 (3.3) | 1 (1.9) | 2 (5.0) | ||
PVA+ NBCA | 2 (2.2) | 1 (1.9) | 1 (2.5) | ||
Angiography only | 6 (6.5) | 3 (5.8) | 3 (7.5) |
Values are presented as number (%).
PVA, polyvinyl alcohol; NBCA, N-butyl cyanoacrylate.
Outcomes of Transcatheter Arterial Embolization
Total (n=86) | Stomach (n=49) | Duodenum (n=37) | P-value | ||
---|---|---|---|---|---|
Technical success | 86 (100.0) | 49 (100.0) | 37 (100.0) | - | |
Rebleeding within 30 days | 19 (22.1) | 13 (26.5) | 6 (16.2) | 0.379 | |
Mortality rate within 30 days | 5 (5.8) | 4 (8.2) | 1 (2.7) | 0.789 | |
Hemorrhage-specific | 4 (4.7) | 3 (6.1) | 1 (2.7) | ||
Underlying disease | 1 (1.2) | 1 (2.0) | 0 (0.0) | ||
Hospital stay (days) | 16.2 (1~67) | 14.2 (1~67) | 18.9 (2~59) | 0.104 | |
Complication | 0.256 | ||||
Ischemic injury of stomach | 1 (1.2) | 1 (2.0) | 0 (0.0) | ||
Partial splenic infarction | 1 (1.2) | 1 (2.0) | 0 (0.0) | ||
Puncture site hematoma | 1 (1.2) | 1 (2.0) | 0 (0.0) |
Values are presented as median (range) or number (%).
Predictors for 30-day Rebleeding after Transcatheter Arterial Embolization
Univariate analysis |
Multivariate analysis |
||||||
---|---|---|---|---|---|---|---|
Rebleeding | No rebleeding | Odds ratio | 95% CI | ||||
Sex, women | 5 (26.3) | 12 (17.9) | 0.420 | ||||
Age (years) | 64.0±13.2 | 67.2±13.6 | 0.362 | ||||
≥2 chronic diseases | 8 (42.1) | 29 (43.3) | 0.927 | ||||
History of peptic ulcer disease | 6 (31.6) | 12 (17.9) | 0.202 | ||||
Antiplatelet agent | 3 (15.8) | 14 (20.9) | 0.623 | ||||
Shock | 16 (84.2) | 39 (58.2) | 0.047 | 2.78 | 0.66~11.68 | 0.162 | |
Coagulopathy | 10 (52.6) | 13 (19.4) | 0.006 | 5.66 | 1.71~18.74 | 0.005 | |
AIMS-65 score, ≥2 high (Ref. 0~1 low) | 18 (94.7) | 52 (77.6) | 0.123 | ||||
Bleeding site, stomach (Ref. duodenum) | 13 (68.4) | 36 (53.7) | 0.260 | ||||
Timing of embolization (Ref. emergent [<12 hours]) | |||||||
Early (12~24 hour) | 3 (15.8) | 17 (25.4) | 0.380 | ||||
Late (>24 hour) | 6 (31.6) | 20 (29.9) | 0.859 | ||||
Empirical embolization (Ref. targeted embolization) | 5 (26.3) | 4 (6.0) | 0.018 | 5.71 | 1.14~28.65 | 0.034 | |
Number of performed embolization (Ref. 1 artery) | |||||||
2 arteries | 0 (0.0) | 6 (9.0) | 0.999 | ||||
≥3 arteries | 2 (10.5) | 1 (1.5) | 0.119 |
Values are presented as mean±standard deviation or number (%).
CI, confidence interval.