- Email: [email protected]
Transcatheter Arterial Embolization of Acute Upper Gastrointestinal Tract Bleeding with N-Butyl-2-Cyanoacrylate
Transcatheter Arterial Embolization of Acute Upper Gastrointestinal Tract Bleeding with N-Butyl-2-Cyanoacrylate Chung-Wei Lee, MD, Kao-Lang Liu, MD, Hsiu-Po Wang, MD, Shyh-Jye Chen, MD, PhD, Yuk-Ming Tsang, MD, and Hon-Man Liu, MD
PURPOSE: To retrospectively analyze outcomes in patients who received an iodized oil formulation of N-butyl-2cyanoacrylate to treat acute nonvariceal upper gastrointestinal tract bleeding. MATERIALS AND METHODS: Sixteen patients with acute upper gastrointestinal tract bleeding underwent angiography and transarterial embolization between April 2004 and December 2005. Patients with negative findings at arteriography and those with lesions in large vessels that did not necessitate a microcatheter for catheterization were excluded. Three types of embolization were used according to the accessible arteries: Localized embolization was used in the bleeding artery, and distal (segmental) or proximal embolization was used in the parent artery according to the final position of the catheter tip. Outcomes, including hemostasis, recurrent bleeding, and complications, were recorded. RESULTS: Immediate hemostasis was achieved in 14 patients (88%). One (6%) patient had technical failure due to arterial dissection, and one (6%) patient had therapeutic failure due to multiple ulcers and coagulopathy. Of the 14 patients with immediate hemostasis, one (7%) had recurrent bleeding 5 days after embolization. Both patients in whom hemostasis was not achieved (12%) died within 1 month of follow-up. Multiple gastric ulcers were seen in two (12%) patients who underwent embolization of a large area, and conservative treatment was sufficient. CONCLUSIONS: The results of this preliminary experience indicate that transarterial embolization with the iodized oil formulation is feasible and effective in the management of nonvariceal upper gastrointestinal tract bleeding; however, a larger number of cases should be investigated. J Vasc Interv Radiol 2007; 18:209 –216 Abbreviations:
NBCA ⫽ N-butyl-2-cyanoacrylate, TAE ⫽ transarterial embolization
TRANSCATHETER management is increasingly common for the treatment of acute upper and lower gastrointestinal tract bleeding. Coils and absorbable gelatin sponge (Gelfoam; Pfizer, New York, N.Y.) are the most
From the Departments of Medical Imaging (C.W.L., K.L.L., S.J.C., Y.M.T., H.M.L.) and Emergency Medicine (H.P.W.), National Taiwan University Hospital and National Taiwan University College of Medicine, No. 7, Chung-Shan South Road, Taipei 100, Taiwan. Received February 16, 2006; final revision received November 7, 2006; accepted December 3, 2006. Address correspondence to K.L.L.; E-mail: [email protected] None of the authors have identified a conflict of interest. © SIR, 2007 DOI: 10.1016/j.jvir.2006.12.003
frequently used embolic agents (1). Tissue adhesive, a liquid embolic agent, offers the advantage of embolization in small vessels (2,3). Intraarterial injection of N-butyl-2-cyanoacrylate (NBCA) has been used with coils to facilitate arterial occlusion and, occasionally, alone to manage bleeding ulcers (1,2). At our institution, NBCA has been used for many years in the embolization of intracranial arteriovenous malformations, with favorable outcomes (4). Because microcoils do not always fit the configuration of the artery to be embolized, the administration of NBCA under adequate control is an ideal alternative in selected cases. Small microcatheters can be used to inject NBCA, and this enhances our
ability to approach the artery with extravasation. The purpose of this study was to retrospectively analyze the outcomes in patients who received NBCA for the management of acute nonvariceal upper gastrointestinal tract bleeding.
MATERIALS AND METHODS This retrospective study was approved by our institutional review board. Informed consent for emergency angiography and embolization was obtained from the patients and/or their families. In four patients (patients 1, 2, 11, and 13), NBCA was not used for the initial embolization procedure because the microcatheters used for our microcoils (0.018 in.)
209
210
•
February 2007
Embolization for Acute Upper GI Bleeding using NBCA
could not enter the bleeding vessel or reach the intended target; however, smaller microcatheters were then used for NBCA injection after communication with the family. When patients and their families did not agree to participate in this study, patients underwent embolization with coils or gelatin sponge particles when extravasation was identified. Patients with negative findings at arteriography were excluded. Patients were also excluded if the bleeding parent artery could be approached by using a 4.1-F catheter (because steel coils could be delivered directly and the additional time and cost of the microcatheter system were not necessary). Between April 2004 and December 2005, 16 patients with acute bleeding in the upper gastrointestinal tract were eligible for this study (Table). There were 11 men and 5 women aged 35– 87 years (mean, 67.7 years). Because 13 patients had massive and acute bleeding at presentation, preprocedural therapeutic panendoscopic images obtained during acute hemorrhage were available in only three patients (patients 4, 8, and 15). Endoscopists failed to stop the bleeding in these three patients before angiography. Extravasation of contrast medium into the gastrointestinal lumen was demonstrated during angiography in 14 of the 16 (88%) patients. The remaining two patients (patients 4 and 14) had localized aneurysmal sacs. The hemorrhage was located in the duodenum in nine patients, in the stomach in five patients, in the jejunum near a gastrojejunostomy in one patient, and in the esophagogastric junction in one patient. One patient with duodenal bleeding (patient 7) had previously undergone embolization with a microcoil at a different location 1 day before he was transferred to our hospital. With regard to the possibility of ischemic injury after the embolization of a large area, the microcatheter was advanced as close to the bleeding vessel as possible in all cases. Microcatheters (Excelsior SL-10, 1.7-F; Boston Scientific, Natick, Mass.) and microguide wires (Agility-10; Cordis, Miami, Fla.) were used in all patients. The microcatheters were navigated into the bleeding artery or the parent arteries of the bleeding vessels. Before
injection of NBCA, a 1-mL syringe was loaded with 0.2– 0.4 mL of NBCA (Histoacryl B; Braun Melsungen, Melsungen, Germany) and iodized oil (Lipiodol; Andre Guerbet, Aulnay-sousBois, France) was added for a total amount of 1 mL (concentration of NBCA, 20%– 40%). The solution was mixed properly by repeated upsidedown movements of the syringe until achieving a homogeneous appearance. This procedure took 1–2 minutes. The field of view and collimation were reduced to cover the area around the catheter tip and the extravasation site, with field of view reduced to about 20 cm and collimation about 10 cm ⫻ 10 cm. Breath holding was not necessary. This enabled better visualization of NBCA coming from the microcatheter because the smaller field of view allowed a larger image, and small collimation eliminated undesired radiation. The microcatheter, including the hub, was flushed with 3–5 mL of 5% dextrose in water to avoid gluing and occlusion of the lumen during injection of the NBCA mixture. Dextrose in water can avoid immediate coagulation of the NBCA. This procedure took 20 –30 seconds. About 0.3 mL of the mixture was rapidly injected into the microcatheter to fill the dead space without fluoroscopy, and it could avoid unnecessary radiation dose. Then, under fluoroscopic guidance, the mixture was slowly and carefully injected for transarterial embolization (TAE). The mixture was continuously injected until it reached our end point, and the microcatheter was removed rapidly and discarded. Each injection took 20 – 60 seconds. Dextrose water was not injected after the administration of the NBCA–iodized oil mixture (the sandwich method). If the bleeding artery could be directly catheterized by using a microcatheter, 30%– 40% of the NBCA mixture was slowly injected until there was minimal regurgitation (Fig 1) or extravasation of the mixture from the bleeding artery. About 0.1 mL of the mixture was injected into the artery to achieve localized embolization (Fig 2a, b). If the parent artery was catheterized but the bleeding vessel itself could not be entered with the microcatheter and when it was possible to reach the location distal to the orifice of the bleeding vessel, the distal
JVIR
portion was occluded with a very slow injection of a higher concentration (40%) of NBCA. Then, the microcatheter was withdrawn to the proximal portion for continuous injection. A total of 0.2– 0.3 mL of the mixture was delivered. This technique was classified as segmental embolization (Fig 2c, d). If the bleeding site could not be approached or the catheter could not reach the distal portion of the orifice of the bleeding artery, 0.2– 0.3 mL of 20%–30% NBCA was injected at a proximal segment, with opacification of the bleeding vessels (patient 13; Fig 3) or occlusion of only the proximal arteries (patients 14 and 15; Fig 4). This technique was classified as proximal embolization (Fig 2e, f). Three patients with bleeding from the duodenum underwent panendoscopy 9 days (patients 1 and 3) or 13 days (patient 5) after TAE, and two patients (patients 10 and 11) with bleeding from the stomach underwent panendoscopy 2 days after TAE. Clinical outcomes were recorded.
RESULTS Fifteen of the 16 patients received one injection of the NBCA–iodized oil mixture to occlude the target arteries. One patient (patient 14; Fig 4) received two injections in different arteries. Immediate hemostasis was achieved in 14 of the 16 (88%) patients. The survival rate at 1 month was 88% (14/16 patients). Nine (56%) patients underwent localized embolization. One patient (patient 7) with a primary lymphoma of the central nervous system died from multiple organ failure (rather than gastrointestinal bleeding) 80 days after TAE. One therapeutic failure occurred (patient 8). In this case, the patient had undergone panendoscopy before TAE, and multiple hemorrhagic duodenal ulcers were seen but not successfully controlled. Although the main extravasation was embolized, the patient had persistent coffee-ground vomit after TAE. Angiography performed 1 and 4 days after TAE did not reveal extravasation of contrast medium. Only diffuse hyperemia of the duodenum was noted at angiography. Despite intraarterial infusion of vasopressin after the third angiographic study, the patient’s general condition declined rapidly (al-
Volume 18
Number 2
Lee et al
•
211
Summary of Patients Treated with Intraarterial NBCA Injection for Acute Upper Gastrointestinal Tract Bleeding Patient No./Sex/Age (Years)
Site of Bleeding
1/F/87
Duodenum
2/F/72
Duodenum
3/F/75 4/M/42 5/F/82
Duodenum Duodenum Duodenum
6/M/80
Duodenum
7/M/73
Duodenum
8/M/80
Duodenum
9/M/39
Jejunum
10/F/54
Stomach
11/M/87
Stomach
12/M/69
Stomach
13/M/62
Duodenum
14/M/75
Stomach
15/M/35
Esophagogastric junction
16/M/71
Stomach
Underlying Disease Diabetes mellitus, hypertension, cerebrovascular accident Cerebrovascular accident, chronic renal failure, pneumonia with respiratory failure Hypertension, diabetes mellitus None End-stage renal disease, coronary artery disease, hypertension Pneumonia with respiratory failure and septic shock, acute renal failure Central nervous system lymphoma, respiratory failure
Findings at Follow-up Panendoscopy
Type of Embolization
Outcome
Localized
Negative
Discharge
Localized
NA
Discharge
Localized Localized Localized
Discharge Discharge No bleeding at 2month follow-up
Localized
Negative NA Negative duodenal ulcer, gastric ulcer scar NA
Localized*
NA
End-stage renal disease, rheumatic heart disease with aortic and mitral valve replacement, bacterial and fungal septic shock, multiple duodenal ulcers at panendoscopy performed before angiography Blunt abdominal trauma after Whipple operation, with leakage and abscess Congestive heart failure, diabetes mellitus, hypertension Pneumonia with respiratory failure Diabetes mellitus, coronary artery disease, common bile duct stone, old cerebrovascular accident Duodenal adenocarcinoma after wedge resection Esophageal ulcers, nephrotic syndrome, congestive heart failure, disseminated tuberculosis Alcoholism and Mallory-Weiss syndrome
Localized
NA
No recurrent bleeding, died from multiple organ failure 80 days later Persistent bleeding, died 5 days later
Localized
NA
No bleeding at 1month follow-up
Segmental
Gastric ulcers
Discharge
Segmental
Gastric ulcers
Discharge
Segmental
NA
No bleeding at 1month follow-up
Proximal
NA
Discharge
Proximal
NA
Recurrent bleeding 5 days later, surgery; discharge
Proximal
NA
History of hepatocellular carcinoma; patient had undergone right lobectomy
Failed
NA
Bacteremia 2 days later, controlled; discharge Surgical ulcer repair; died from hepatic failure 10 days later
No bleeding at 1month follow-up
NA ⫽ not applicable. * The patient had undergone coil embolization at another institution 1 day earlier.
though the amount of coffee-ground vomit had decreased). His family refused further aggressive treatment, and the patient died 5 days after TAE. Three patients (19%) underwent segmental embolization; all three were
discharged without clinical complications. Three patients (19%) underwent proximal embolization. One patient (patient 14) developed repeat bleeding 5 days after embolization and under-
went gastrectomy without complications. One patient with Mallory-Weiss syndrome (patient 15) developed afebrile Klebsiella pneumonia bacteremia 2 days after TAE. This patient was treated with antibiotics for 2 weeks.
212
•
Embolization for Acute Upper GI Bleeding using NBCA
February 2007
JVIR
Figure 1. Images in an 87-year-old woman (patient 1) who presented in shock with massive hematemesis and bloody stool. (a) Celiac arteriogram demonstrates diffuse vasoconstriction due to the patient’s shock. Findings are suggestive of a pseudoaneurysm (arrow), and there is extravasation of contrast medium in the territory of the gastroduodenal artery. Selective arteriography (not shown), however, showed that the bleeding vessel was from the proximal right gastric artery via the proper hepatic artery. (b) Arteriogram obtained after the injection of NBCA into the right gastric artery with localized embolization. The feeding artery (arrow) and pseudoaneurysm (arrowhead) are opacified. (c) Arteriogram obtained after embolization shows that the right gastric artery (arrow) and pseudoaneurysm (arrowhead) are occluded. Endoscopy was performed 9 days after embolization; no definite mucosal lesion was seen.
Because the patient had undergone other procedures, including injection of epinephrine at panendoscopy and insertion of a central venous catheter, the afebrile bacteremia could not be definitely related to our procedure. In four patients (patients 1, 2, 11, and 13), smaller microcatheters (Excelsior SL-10, 1.7-F; Boston Scientific) had to be used because the larger microcatheters could not be advanced far enough for embolization. Two of the four patients underwent local embolization, and one patient each underwent segmental and proximal embolization. Hemostasis was achieved in all four patients. Hemorrhage had not recurred at 1to 20-month follow-up in 13 of the 14 patients in whom immediate hemostasis was achieved. None of the patients
presented with symptoms or signs of gastrointestinal ischemia or clinically significant gastrointestinal stenosis or obstruction. One technical failure occurred (patient 16). This patient had undergone right hepatic lobectomy due to hepatocellular carcinoma and developed hepatic failure. In this case, arterial dissection with total occlusion occurred in the parent artery of the bleeding vessel, a branch of the left gastric artery. Tiny collateral arteries were identified from the greater curvature of the stomach (greater omental artery via the gastroduodenal artery), and these arteries caused persistent extravasation of contrast medium. We tried to approach these arteries for a while but failed even to reach the parent artery of these small collateral ves-
sels. We determined that a further attempt to catheterize these vessels would require a long period of time and worsen the outcome of this patient with shock. Embolization was not performed, and the patient underwent emergency laparotomy for ulcer repair but died from hepatic failure 10 days later. Follow-up panendoscopy performed in three patients with duodenal bleeding (patients 1, 3, and 5) did not reveal any mucosal lesions in the duodenum. Panendoscopy performed 2 days after TAE in the two patients with gastric bleeding (patients 10 and 11) showed multiple gastric ulcers, which could be due to the segmental embolization performed in the parent arteries at the lesser curvature.
Volume 18
Number 2
Lee et al
•
213
Figure 2. Diagrams show the three different locations for microcatheter placement (in green) and the sites of NBCA embolization (in blue). (a) The bleeding artery is catheterized directly and the microcatheter delivered near the bleeding vessel (arrow in a, c, and e). (b) Then, about 0.1 mL of NBCA is injected for localized embolization. (c) If the bleeding artery cannot be directly catheterized and the microcatheter is delivered into the distal portion of the parent artery, (d) about 0.2– 0.3 mL of NBCA is injected for segmental embolization. This ensures that the bleeding vessel will not be reperfused from the distal parent artery. (e) If the bleeding artery cannot be directly catheterized and the microcatheter is delivered only into the proximal portion of the parent artery, about 0.2– 0.3 mL of NBCA is injected for proximal embolization in (f) ideal or (g) actual conditions.
DISCUSSION Although cyanoacrylates are useful for controlling hemorrhage, complications due to bowel ischemia have been reported (3). However, NBCA, the most frequently used tissue adhesive, is increasingly used in the endoscopic treatment of variceal or nonvariceal upper gastrointestinal tract bleeding (5–7). Modern endovascular devices, which have smaller outer diameters, easier flexibility, and better maneuverability, enable the more precise assessment of target vessels. Delivery of NBCA can be restricted to a limited area, and extensive ischemia can be avoided when caution is used. The use of NBCA in addition to coils is also effective in the embolization of bleeding ulcers (2). The precise delivery of liquid embolic material requires that the interventional radiologist be highly skilled.
The learning curve for liquid embolic material is higher than that for coils or particles. Inadequate control of NBCA may result in extensive penetration into the microcirculation and regurgitation into undesired targets. This prohibits the extensive use of this agent all over the world. Long-term experiences with NBCA, however, have been established at our institution, and we are comfortable with the use of this embolic agent when conditions are suitable. N-Butyl-2-cyanoacrylate can be injected through a small microcatheter and, with experience, the amount to be delivered can be precisely controlled. With use of the smaller microcatheter, we successfully catheterized the target arteries that were not successfully catheterized with larger microcatheters, and these patients (n ⫽ 4) had favorable outcomes. Immediate hemostasis was achieved
in 14 of the 16 (88%) patients. One of the 14 (7%) patients had recurrent bleeding 5 days after the initial TAE. There was one technical failure and one therapeutic failure, both of which resulted in the death of the patient. The case of technical failure was not directly related to NBCA but to microcatheter manipulation. Because we initially tried to use NBCA to manage the patient’s bleeding, however, the parent artery was dissected and this case was determined to be a technique failure. The patient in whom immediate hemostasis was not achieved (patient 8) had multiple duodenal ulcers with hemorrhage, as observed at panendoscopy. Coagulopathy was also found, and this patient’s sepsis could not be controlled. Therefore, although angiographically demonstrable hemostasis could be achieved, the procedure was classified as a clinical failure, presum-
214
•
Embolization for Acute Upper GI Bleeding using NBCA
February 2007
JVIR
Figure 3. Images in a 62-year-old man (patient 13) who underwent wedge resection of the second portion of the duodenum to remove a primary adenocarcinoma. There was a massive amount of blood in the stool and he had shock 1 week after surgery. (a) Superior mesenteric arteriogram shows a pseudoaneurysm (arrow) from a branch of the inferior pancreaticoduodenal artery. (b) Arteriogram obtained after proximal embolization of the inferior pancreaticoduodenal artery with NBCA in the embolized arteries (arrows) shows that flow from the superior mesenteric artery is blocked. (c) Celiac arteriogram obtained after embolization shows that there is no collateral supply. The patient recovered well and went home 1 week after embolization.
ably due to persistent angiographically undetectable bleeding from the ulcers. The procedure time is dependent on the manipulation of the microcatheter and microguide wire rather than the deployment of the microcoil or the injection of NBCA. Fourteen of the 15 patients who underwent NBCA embolization had only one injection. It took less than 5 minutes to perform each injection when the microcatheter reached the target. In addition, the cost of dextrose water ($1.50), NBCA ($45.50), and iodized
oil ($6.00) is less than that of a microcoil ($84.80) and a coil pusher ($140.00). Because only one injection was necessary in most patients, the cost of NBCA embolization was cheaper than that with microcoils and coil pushers. Attempts to advance microcatheters to the source of hemorrhage proved difficult in many of the cases discussed herein due to the small size of the feeding vessels. Vasoconstriction compounded the problem in patients with hypotension. We found that we were able to enter these arte-
rial feeding vessels by using even smaller microcatheters. No technical difficulties were encountered in the nine patients who underwent localized embolization. N-Butyl-2-cyanoacrylate–iodized oil mixture from the microcatheter tip was monitored with fluoroscopy. When this material was identified, slow injection was continued until minimal regurgitation of the embolic agent was seen. Four injections were performed by radiologists who had never used NBCA before, although they were supervised by experienced radiologists. With ba-
Volume 18
Number 2
Lee et al
•
215
Figure 4. Images in a 75-year-old man (patient 14) with disseminated tuberculosis, multiple esophageal ulcers, and severe coagulopathy who developed massive hematemesis. (a) Celiac arteriogram shows an aneurysmal sac (*) at the gastric fundus with blood supply from the left gastric artery (arrow) and short gastric artery (arrowhead). (b) Arteriogram obtained after the injection of NBCA, which could be injected only at the proximal portion of the left gastric artery because of the small and torturous artery. NBCA (arrow and arrowhead) could not reach the pseudoaneurysm from either side. Although immediate hemostasis was achieved and the patient’s vital signs stabilized, hemorrhage recurred 5 days later.
sic training in NBCA use, most interventional radiologists may find that the technique is not so difficult. Segmental and proximal embolization with NBCA was performed in six patients (seven injections) by experienced radiologists. We believe that the technique might not be performed well by all interventional radiologists. If localized embolization is not possible, however, this does not exclude the feasibility of embolization with microcoils or particles as long as suitable types are available. Five of the six patients with extravasation in the stomach or esophagogastric junction were treated with segmental (n ⫽ 3) or proximal (n ⫽ 2) embolization. By comparison, localized embolization was successfully achieved in eight of nine patients with extravasation in the duodenum or jejunum. More complex arterial network and arterial angulation might be the cause of this difference. Bacteremia was found in one patient (patient 15). Although bacteremia has been reported after the endoscopic injection of NBCA (8), to our knowledge it has not been reported after intraarterial injection. It could be difficult to determine the true cause of bacteremia in this patient because he had undergone other invasive procedures. The entity of bleeding in this patient was different from that of
other patients. The patient had Mallory-Weiss syndrome, and the mucosal laceration may have provided a route for bacteria to enter after embolization. Other procedures, however, may also have caused bacteremia. Multiple ulcers were observed in the two patients who underwent panendoscopy 2 days after segmental TAE. Conservative treatment was sufficient in these patients. The ulcers could have been the result of mucosal ischemia after large areas were treated with TAE. Similar results have been observed after TAE with small particles (9). Multiple ulcers, however, are not uncommon in our daily practice, and the patients could have had multiple ulcers at the time of massive bleeding. We were limited in demonstrating the relationship between NBCA embolization and the resultant ulcers because baseline and follow-up panendoscopic studies were not available in most patients. This probable complication, however, was minor compared with life-threatening hemorrhage (10). One patient (patient 13) underwent proximal embolization at the inferior pancreaticoduodenal artery via the superior mesenteric artery to treat pseudoaneurysmal bleeding at the third portion of the duodenum. This case demonstrated the potential advantage of using NBCA as an embolic
agent (Fig 3). Because the microcatheter could not be advanced to the bleeding vessel, the injection was performed proximally. The bleeding artery and the parent artery were both completely occluded. If the parent artery (inferior pancreaticoduodenal artery) on each side of the bleeding vessel had not been embolized, the hemorrhage may have persisted owing to the collateral supply. In this instance, NBCA would likely have been more effective than microcoils unless embolization of different sites could be accomplished. A duodenal mucosal lesion was not demonstrated at panendoscopy in the three patients with duodenal bleeding 9 and 13 days after selective TAE. Acute ischemic insults occurring shortly after TAE may have been missed at the delayed panendoscopic examinations. Although no long-term mucosal complications were encountered in our series, the number of patients with endoscopic evaluation was limited. Further evaluation with more patients is indicated. N-Butyl-2-cyanoacrylate, however, may be a potential alternative to other embolic agents. In conclusion, a small microcatheter can be used with liquid embolic agent to access bleeding vessels. The length and caliber of a bleeding artery, which might cause coils or microcoils to be improperly positioned, do not hamper
216
•
February 2007
Embolization for Acute Upper GI Bleeding using NBCA
embolization with NBCA. When performed by experienced interventional radiologists, the injection of NBCA appears to be a safe alternative to microcoils and embolic particles in the treatment of acute upper gastrointestinal tract bleeding. If the catheters can be placed as close as possible to the site of extravasation, the outcome after NBCA injection is favorable. Even in some cases in which the bleeding artery cannot be catheterized, NBCA is a good alternative to microcoils and embolic particles. In our preliminary experience, TAE with the iodized oil formulation was effective for managing nonvariceal upper gastrointestinal tract bleeding. More cases and experience are needed to verify these findings, and the long-term complications should be monitored. Acknowledgments: This work was supported by a grant from the Department of Medical Research, National Taiwan University Hospital (NTUH095-000382). The
authors acknowledge Dr. Kathy Crabtree for her detailed revision of our article. References 1. Kramer SC, Gorich J, Rilinger N, et al. Embolization for gastrointestinal hemorrhages. Eur Radiol 2000; 10:802– 805. 2. Toyoda H, Nakano S, Kumada T, et al. Estimation of usefulness of N-butyl-2-cyanoacrylate–lipiodol mixture in transcatheter arterial embolization for urgent control of life-threatening massive bleeding from gastric or duodenal ulcer. J Gastroenterol Hepatol 1996; 11: 252–258. 3. Goldman ML, Freeny PC, Tallman JM, et al. Transcatheter vascular occlusion therapy with isobutyl 2-cyanoacrylate (bucrylate) for control of massive upper-gastrointestinal bleeding. Radiology 1978; 129:41– 49. 4. Liu HM, Huang YC, Wang YH. Embolization of cerebral arteriovenous malformations with N-butyl-2-cyanoacrylate. J Formos Med Assoc 2000; 99: 906 –913. 5. Lee KJ, Kim JH, Hahm KB, Cho SW, Park YS. Randomized trial of N-butyl-2-cyanoacrylate compared with in-
6.
7.
8.
9.
10.
JVIR
jection of hypertonic saline-epinephrine in the endoscopic treatment of bleeding peptic ulcers. Endoscopy 2000; 32:505–511. Yok KY, Kum CK, Goh PM. Endoscopic hemostasis of upper gastrointestinal bleeding with histoacryl: last resort before surgery. Endoscopy 1996; 28:256 –258. Kish JW, Katz MD, Marx MV, Harrell DS, Hanks SE. N-butyl cyanoacrylate embolization for control of acute arterial hemorrhage. J Vasc Interv Radiol 2004; 15:689 – 695. Wahl P, Lammer F, Conen D, Schlumpf R, Bock A. Septic complications after injection of N-butyl-2-cyanoacrylate: report of two cases and review. Gastrointest Endosc 2004; 59:911–916. Rosch J, Keller FS, Kozak B, Niles N, Dotter CT. Gelfoam powder embolization of the left gastric artery in treatment of massive small-vessel gastric bleeding. Radiology 1984; 151:365–370. Bandi R, Shetty PC, Sharma RP. Superselective arterial embolization for the treatment of lower gastrointestinal hemorrhage. J Vasc Interv Radiol 2001; 12:1399 –1405.
PURPOSE: To retrospectively analyze outcomes in patients who received an iodized oil formulation of N-butyl-2cyanoacrylate to treat acute nonvariceal upper gastrointestinal tract bleeding. MATERIALS AND METHODS: Sixteen patients with acute upper gastrointestinal tract bleeding underwent angiography and transarterial embolization between April 2004 and December 2005. Patients with negative findings at arteriography and those with lesions in large vessels that did not necessitate a microcatheter for catheterization were excluded. Three types of embolization were used according to the accessible arteries: Localized embolization was used in the bleeding artery, and distal (segmental) or proximal embolization was used in the parent artery according to the final position of the catheter tip. Outcomes, including hemostasis, recurrent bleeding, and complications, were recorded. RESULTS: Immediate hemostasis was achieved in 14 patients (88%). One (6%) patient had technical failure due to arterial dissection, and one (6%) patient had therapeutic failure due to multiple ulcers and coagulopathy. Of the 14 patients with immediate hemostasis, one (7%) had recurrent bleeding 5 days after embolization. Both patients in whom hemostasis was not achieved (12%) died within 1 month of follow-up. Multiple gastric ulcers were seen in two (12%) patients who underwent embolization of a large area, and conservative treatment was sufficient. CONCLUSIONS: The results of this preliminary experience indicate that transarterial embolization with the iodized oil formulation is feasible and effective in the management of nonvariceal upper gastrointestinal tract bleeding; however, a larger number of cases should be investigated. J Vasc Interv Radiol 2007; 18:209 –216 Abbreviations:
NBCA ⫽ N-butyl-2-cyanoacrylate, TAE ⫽ transarterial embolization
TRANSCATHETER management is increasingly common for the treatment of acute upper and lower gastrointestinal tract bleeding. Coils and absorbable gelatin sponge (Gelfoam; Pfizer, New York, N.Y.) are the most
From the Departments of Medical Imaging (C.W.L., K.L.L., S.J.C., Y.M.T., H.M.L.) and Emergency Medicine (H.P.W.), National Taiwan University Hospital and National Taiwan University College of Medicine, No. 7, Chung-Shan South Road, Taipei 100, Taiwan. Received February 16, 2006; final revision received November 7, 2006; accepted December 3, 2006. Address correspondence to K.L.L.; E-mail: [email protected] None of the authors have identified a conflict of interest. © SIR, 2007 DOI: 10.1016/j.jvir.2006.12.003
frequently used embolic agents (1). Tissue adhesive, a liquid embolic agent, offers the advantage of embolization in small vessels (2,3). Intraarterial injection of N-butyl-2-cyanoacrylate (NBCA) has been used with coils to facilitate arterial occlusion and, occasionally, alone to manage bleeding ulcers (1,2). At our institution, NBCA has been used for many years in the embolization of intracranial arteriovenous malformations, with favorable outcomes (4). Because microcoils do not always fit the configuration of the artery to be embolized, the administration of NBCA under adequate control is an ideal alternative in selected cases. Small microcatheters can be used to inject NBCA, and this enhances our
ability to approach the artery with extravasation. The purpose of this study was to retrospectively analyze the outcomes in patients who received NBCA for the management of acute nonvariceal upper gastrointestinal tract bleeding.
MATERIALS AND METHODS This retrospective study was approved by our institutional review board. Informed consent for emergency angiography and embolization was obtained from the patients and/or their families. In four patients (patients 1, 2, 11, and 13), NBCA was not used for the initial embolization procedure because the microcatheters used for our microcoils (0.018 in.)
209
210
•
February 2007
Embolization for Acute Upper GI Bleeding using NBCA
could not enter the bleeding vessel or reach the intended target; however, smaller microcatheters were then used for NBCA injection after communication with the family. When patients and their families did not agree to participate in this study, patients underwent embolization with coils or gelatin sponge particles when extravasation was identified. Patients with negative findings at arteriography were excluded. Patients were also excluded if the bleeding parent artery could be approached by using a 4.1-F catheter (because steel coils could be delivered directly and the additional time and cost of the microcatheter system were not necessary). Between April 2004 and December 2005, 16 patients with acute bleeding in the upper gastrointestinal tract were eligible for this study (Table). There were 11 men and 5 women aged 35– 87 years (mean, 67.7 years). Because 13 patients had massive and acute bleeding at presentation, preprocedural therapeutic panendoscopic images obtained during acute hemorrhage were available in only three patients (patients 4, 8, and 15). Endoscopists failed to stop the bleeding in these three patients before angiography. Extravasation of contrast medium into the gastrointestinal lumen was demonstrated during angiography in 14 of the 16 (88%) patients. The remaining two patients (patients 4 and 14) had localized aneurysmal sacs. The hemorrhage was located in the duodenum in nine patients, in the stomach in five patients, in the jejunum near a gastrojejunostomy in one patient, and in the esophagogastric junction in one patient. One patient with duodenal bleeding (patient 7) had previously undergone embolization with a microcoil at a different location 1 day before he was transferred to our hospital. With regard to the possibility of ischemic injury after the embolization of a large area, the microcatheter was advanced as close to the bleeding vessel as possible in all cases. Microcatheters (Excelsior SL-10, 1.7-F; Boston Scientific, Natick, Mass.) and microguide wires (Agility-10; Cordis, Miami, Fla.) were used in all patients. The microcatheters were navigated into the bleeding artery or the parent arteries of the bleeding vessels. Before
injection of NBCA, a 1-mL syringe was loaded with 0.2– 0.4 mL of NBCA (Histoacryl B; Braun Melsungen, Melsungen, Germany) and iodized oil (Lipiodol; Andre Guerbet, Aulnay-sousBois, France) was added for a total amount of 1 mL (concentration of NBCA, 20%– 40%). The solution was mixed properly by repeated upsidedown movements of the syringe until achieving a homogeneous appearance. This procedure took 1–2 minutes. The field of view and collimation were reduced to cover the area around the catheter tip and the extravasation site, with field of view reduced to about 20 cm and collimation about 10 cm ⫻ 10 cm. Breath holding was not necessary. This enabled better visualization of NBCA coming from the microcatheter because the smaller field of view allowed a larger image, and small collimation eliminated undesired radiation. The microcatheter, including the hub, was flushed with 3–5 mL of 5% dextrose in water to avoid gluing and occlusion of the lumen during injection of the NBCA mixture. Dextrose in water can avoid immediate coagulation of the NBCA. This procedure took 20 –30 seconds. About 0.3 mL of the mixture was rapidly injected into the microcatheter to fill the dead space without fluoroscopy, and it could avoid unnecessary radiation dose. Then, under fluoroscopic guidance, the mixture was slowly and carefully injected for transarterial embolization (TAE). The mixture was continuously injected until it reached our end point, and the microcatheter was removed rapidly and discarded. Each injection took 20 – 60 seconds. Dextrose water was not injected after the administration of the NBCA–iodized oil mixture (the sandwich method). If the bleeding artery could be directly catheterized by using a microcatheter, 30%– 40% of the NBCA mixture was slowly injected until there was minimal regurgitation (Fig 1) or extravasation of the mixture from the bleeding artery. About 0.1 mL of the mixture was injected into the artery to achieve localized embolization (Fig 2a, b). If the parent artery was catheterized but the bleeding vessel itself could not be entered with the microcatheter and when it was possible to reach the location distal to the orifice of the bleeding vessel, the distal
JVIR
portion was occluded with a very slow injection of a higher concentration (40%) of NBCA. Then, the microcatheter was withdrawn to the proximal portion for continuous injection. A total of 0.2– 0.3 mL of the mixture was delivered. This technique was classified as segmental embolization (Fig 2c, d). If the bleeding site could not be approached or the catheter could not reach the distal portion of the orifice of the bleeding artery, 0.2– 0.3 mL of 20%–30% NBCA was injected at a proximal segment, with opacification of the bleeding vessels (patient 13; Fig 3) or occlusion of only the proximal arteries (patients 14 and 15; Fig 4). This technique was classified as proximal embolization (Fig 2e, f). Three patients with bleeding from the duodenum underwent panendoscopy 9 days (patients 1 and 3) or 13 days (patient 5) after TAE, and two patients (patients 10 and 11) with bleeding from the stomach underwent panendoscopy 2 days after TAE. Clinical outcomes were recorded.
RESULTS Fifteen of the 16 patients received one injection of the NBCA–iodized oil mixture to occlude the target arteries. One patient (patient 14; Fig 4) received two injections in different arteries. Immediate hemostasis was achieved in 14 of the 16 (88%) patients. The survival rate at 1 month was 88% (14/16 patients). Nine (56%) patients underwent localized embolization. One patient (patient 7) with a primary lymphoma of the central nervous system died from multiple organ failure (rather than gastrointestinal bleeding) 80 days after TAE. One therapeutic failure occurred (patient 8). In this case, the patient had undergone panendoscopy before TAE, and multiple hemorrhagic duodenal ulcers were seen but not successfully controlled. Although the main extravasation was embolized, the patient had persistent coffee-ground vomit after TAE. Angiography performed 1 and 4 days after TAE did not reveal extravasation of contrast medium. Only diffuse hyperemia of the duodenum was noted at angiography. Despite intraarterial infusion of vasopressin after the third angiographic study, the patient’s general condition declined rapidly (al-
Volume 18
Number 2
Lee et al
•
211
Summary of Patients Treated with Intraarterial NBCA Injection for Acute Upper Gastrointestinal Tract Bleeding Patient No./Sex/Age (Years)
Site of Bleeding
1/F/87
Duodenum
2/F/72
Duodenum
3/F/75 4/M/42 5/F/82
Duodenum Duodenum Duodenum
6/M/80
Duodenum
7/M/73
Duodenum
8/M/80
Duodenum
9/M/39
Jejunum
10/F/54
Stomach
11/M/87
Stomach
12/M/69
Stomach
13/M/62
Duodenum
14/M/75
Stomach
15/M/35
Esophagogastric junction
16/M/71
Stomach
Underlying Disease Diabetes mellitus, hypertension, cerebrovascular accident Cerebrovascular accident, chronic renal failure, pneumonia with respiratory failure Hypertension, diabetes mellitus None End-stage renal disease, coronary artery disease, hypertension Pneumonia with respiratory failure and septic shock, acute renal failure Central nervous system lymphoma, respiratory failure
Findings at Follow-up Panendoscopy
Type of Embolization
Outcome
Localized
Negative
Discharge
Localized
NA
Discharge
Localized Localized Localized
Discharge Discharge No bleeding at 2month follow-up
Localized
Negative NA Negative duodenal ulcer, gastric ulcer scar NA
Localized*
NA
End-stage renal disease, rheumatic heart disease with aortic and mitral valve replacement, bacterial and fungal septic shock, multiple duodenal ulcers at panendoscopy performed before angiography Blunt abdominal trauma after Whipple operation, with leakage and abscess Congestive heart failure, diabetes mellitus, hypertension Pneumonia with respiratory failure Diabetes mellitus, coronary artery disease, common bile duct stone, old cerebrovascular accident Duodenal adenocarcinoma after wedge resection Esophageal ulcers, nephrotic syndrome, congestive heart failure, disseminated tuberculosis Alcoholism and Mallory-Weiss syndrome
Localized
NA
No recurrent bleeding, died from multiple organ failure 80 days later Persistent bleeding, died 5 days later
Localized
NA
No bleeding at 1month follow-up
Segmental
Gastric ulcers
Discharge
Segmental
Gastric ulcers
Discharge
Segmental
NA
No bleeding at 1month follow-up
Proximal
NA
Discharge
Proximal
NA
Recurrent bleeding 5 days later, surgery; discharge
Proximal
NA
History of hepatocellular carcinoma; patient had undergone right lobectomy
Failed
NA
Bacteremia 2 days later, controlled; discharge Surgical ulcer repair; died from hepatic failure 10 days later
No bleeding at 1month follow-up
NA ⫽ not applicable. * The patient had undergone coil embolization at another institution 1 day earlier.
though the amount of coffee-ground vomit had decreased). His family refused further aggressive treatment, and the patient died 5 days after TAE. Three patients (19%) underwent segmental embolization; all three were
discharged without clinical complications. Three patients (19%) underwent proximal embolization. One patient (patient 14) developed repeat bleeding 5 days after embolization and under-
went gastrectomy without complications. One patient with Mallory-Weiss syndrome (patient 15) developed afebrile Klebsiella pneumonia bacteremia 2 days after TAE. This patient was treated with antibiotics for 2 weeks.
212
•
Embolization for Acute Upper GI Bleeding using NBCA
February 2007
JVIR
Figure 1. Images in an 87-year-old woman (patient 1) who presented in shock with massive hematemesis and bloody stool. (a) Celiac arteriogram demonstrates diffuse vasoconstriction due to the patient’s shock. Findings are suggestive of a pseudoaneurysm (arrow), and there is extravasation of contrast medium in the territory of the gastroduodenal artery. Selective arteriography (not shown), however, showed that the bleeding vessel was from the proximal right gastric artery via the proper hepatic artery. (b) Arteriogram obtained after the injection of NBCA into the right gastric artery with localized embolization. The feeding artery (arrow) and pseudoaneurysm (arrowhead) are opacified. (c) Arteriogram obtained after embolization shows that the right gastric artery (arrow) and pseudoaneurysm (arrowhead) are occluded. Endoscopy was performed 9 days after embolization; no definite mucosal lesion was seen.
Because the patient had undergone other procedures, including injection of epinephrine at panendoscopy and insertion of a central venous catheter, the afebrile bacteremia could not be definitely related to our procedure. In four patients (patients 1, 2, 11, and 13), smaller microcatheters (Excelsior SL-10, 1.7-F; Boston Scientific) had to be used because the larger microcatheters could not be advanced far enough for embolization. Two of the four patients underwent local embolization, and one patient each underwent segmental and proximal embolization. Hemostasis was achieved in all four patients. Hemorrhage had not recurred at 1to 20-month follow-up in 13 of the 14 patients in whom immediate hemostasis was achieved. None of the patients
presented with symptoms or signs of gastrointestinal ischemia or clinically significant gastrointestinal stenosis or obstruction. One technical failure occurred (patient 16). This patient had undergone right hepatic lobectomy due to hepatocellular carcinoma and developed hepatic failure. In this case, arterial dissection with total occlusion occurred in the parent artery of the bleeding vessel, a branch of the left gastric artery. Tiny collateral arteries were identified from the greater curvature of the stomach (greater omental artery via the gastroduodenal artery), and these arteries caused persistent extravasation of contrast medium. We tried to approach these arteries for a while but failed even to reach the parent artery of these small collateral ves-
sels. We determined that a further attempt to catheterize these vessels would require a long period of time and worsen the outcome of this patient with shock. Embolization was not performed, and the patient underwent emergency laparotomy for ulcer repair but died from hepatic failure 10 days later. Follow-up panendoscopy performed in three patients with duodenal bleeding (patients 1, 3, and 5) did not reveal any mucosal lesions in the duodenum. Panendoscopy performed 2 days after TAE in the two patients with gastric bleeding (patients 10 and 11) showed multiple gastric ulcers, which could be due to the segmental embolization performed in the parent arteries at the lesser curvature.
Volume 18
Number 2
Lee et al
•
213
Figure 2. Diagrams show the three different locations for microcatheter placement (in green) and the sites of NBCA embolization (in blue). (a) The bleeding artery is catheterized directly and the microcatheter delivered near the bleeding vessel (arrow in a, c, and e). (b) Then, about 0.1 mL of NBCA is injected for localized embolization. (c) If the bleeding artery cannot be directly catheterized and the microcatheter is delivered into the distal portion of the parent artery, (d) about 0.2– 0.3 mL of NBCA is injected for segmental embolization. This ensures that the bleeding vessel will not be reperfused from the distal parent artery. (e) If the bleeding artery cannot be directly catheterized and the microcatheter is delivered only into the proximal portion of the parent artery, about 0.2– 0.3 mL of NBCA is injected for proximal embolization in (f) ideal or (g) actual conditions.
DISCUSSION Although cyanoacrylates are useful for controlling hemorrhage, complications due to bowel ischemia have been reported (3). However, NBCA, the most frequently used tissue adhesive, is increasingly used in the endoscopic treatment of variceal or nonvariceal upper gastrointestinal tract bleeding (5–7). Modern endovascular devices, which have smaller outer diameters, easier flexibility, and better maneuverability, enable the more precise assessment of target vessels. Delivery of NBCA can be restricted to a limited area, and extensive ischemia can be avoided when caution is used. The use of NBCA in addition to coils is also effective in the embolization of bleeding ulcers (2). The precise delivery of liquid embolic material requires that the interventional radiologist be highly skilled.
The learning curve for liquid embolic material is higher than that for coils or particles. Inadequate control of NBCA may result in extensive penetration into the microcirculation and regurgitation into undesired targets. This prohibits the extensive use of this agent all over the world. Long-term experiences with NBCA, however, have been established at our institution, and we are comfortable with the use of this embolic agent when conditions are suitable. N-Butyl-2-cyanoacrylate can be injected through a small microcatheter and, with experience, the amount to be delivered can be precisely controlled. With use of the smaller microcatheter, we successfully catheterized the target arteries that were not successfully catheterized with larger microcatheters, and these patients (n ⫽ 4) had favorable outcomes. Immediate hemostasis was achieved
in 14 of the 16 (88%) patients. One of the 14 (7%) patients had recurrent bleeding 5 days after the initial TAE. There was one technical failure and one therapeutic failure, both of which resulted in the death of the patient. The case of technical failure was not directly related to NBCA but to microcatheter manipulation. Because we initially tried to use NBCA to manage the patient’s bleeding, however, the parent artery was dissected and this case was determined to be a technique failure. The patient in whom immediate hemostasis was not achieved (patient 8) had multiple duodenal ulcers with hemorrhage, as observed at panendoscopy. Coagulopathy was also found, and this patient’s sepsis could not be controlled. Therefore, although angiographically demonstrable hemostasis could be achieved, the procedure was classified as a clinical failure, presum-
214
•
Embolization for Acute Upper GI Bleeding using NBCA
February 2007
JVIR
Figure 3. Images in a 62-year-old man (patient 13) who underwent wedge resection of the second portion of the duodenum to remove a primary adenocarcinoma. There was a massive amount of blood in the stool and he had shock 1 week after surgery. (a) Superior mesenteric arteriogram shows a pseudoaneurysm (arrow) from a branch of the inferior pancreaticoduodenal artery. (b) Arteriogram obtained after proximal embolization of the inferior pancreaticoduodenal artery with NBCA in the embolized arteries (arrows) shows that flow from the superior mesenteric artery is blocked. (c) Celiac arteriogram obtained after embolization shows that there is no collateral supply. The patient recovered well and went home 1 week after embolization.
ably due to persistent angiographically undetectable bleeding from the ulcers. The procedure time is dependent on the manipulation of the microcatheter and microguide wire rather than the deployment of the microcoil or the injection of NBCA. Fourteen of the 15 patients who underwent NBCA embolization had only one injection. It took less than 5 minutes to perform each injection when the microcatheter reached the target. In addition, the cost of dextrose water ($1.50), NBCA ($45.50), and iodized
oil ($6.00) is less than that of a microcoil ($84.80) and a coil pusher ($140.00). Because only one injection was necessary in most patients, the cost of NBCA embolization was cheaper than that with microcoils and coil pushers. Attempts to advance microcatheters to the source of hemorrhage proved difficult in many of the cases discussed herein due to the small size of the feeding vessels. Vasoconstriction compounded the problem in patients with hypotension. We found that we were able to enter these arte-
rial feeding vessels by using even smaller microcatheters. No technical difficulties were encountered in the nine patients who underwent localized embolization. N-Butyl-2-cyanoacrylate–iodized oil mixture from the microcatheter tip was monitored with fluoroscopy. When this material was identified, slow injection was continued until minimal regurgitation of the embolic agent was seen. Four injections were performed by radiologists who had never used NBCA before, although they were supervised by experienced radiologists. With ba-
Volume 18
Number 2
Lee et al
•
215
Figure 4. Images in a 75-year-old man (patient 14) with disseminated tuberculosis, multiple esophageal ulcers, and severe coagulopathy who developed massive hematemesis. (a) Celiac arteriogram shows an aneurysmal sac (*) at the gastric fundus with blood supply from the left gastric artery (arrow) and short gastric artery (arrowhead). (b) Arteriogram obtained after the injection of NBCA, which could be injected only at the proximal portion of the left gastric artery because of the small and torturous artery. NBCA (arrow and arrowhead) could not reach the pseudoaneurysm from either side. Although immediate hemostasis was achieved and the patient’s vital signs stabilized, hemorrhage recurred 5 days later.
sic training in NBCA use, most interventional radiologists may find that the technique is not so difficult. Segmental and proximal embolization with NBCA was performed in six patients (seven injections) by experienced radiologists. We believe that the technique might not be performed well by all interventional radiologists. If localized embolization is not possible, however, this does not exclude the feasibility of embolization with microcoils or particles as long as suitable types are available. Five of the six patients with extravasation in the stomach or esophagogastric junction were treated with segmental (n ⫽ 3) or proximal (n ⫽ 2) embolization. By comparison, localized embolization was successfully achieved in eight of nine patients with extravasation in the duodenum or jejunum. More complex arterial network and arterial angulation might be the cause of this difference. Bacteremia was found in one patient (patient 15). Although bacteremia has been reported after the endoscopic injection of NBCA (8), to our knowledge it has not been reported after intraarterial injection. It could be difficult to determine the true cause of bacteremia in this patient because he had undergone other invasive procedures. The entity of bleeding in this patient was different from that of
other patients. The patient had Mallory-Weiss syndrome, and the mucosal laceration may have provided a route for bacteria to enter after embolization. Other procedures, however, may also have caused bacteremia. Multiple ulcers were observed in the two patients who underwent panendoscopy 2 days after segmental TAE. Conservative treatment was sufficient in these patients. The ulcers could have been the result of mucosal ischemia after large areas were treated with TAE. Similar results have been observed after TAE with small particles (9). Multiple ulcers, however, are not uncommon in our daily practice, and the patients could have had multiple ulcers at the time of massive bleeding. We were limited in demonstrating the relationship between NBCA embolization and the resultant ulcers because baseline and follow-up panendoscopic studies were not available in most patients. This probable complication, however, was minor compared with life-threatening hemorrhage (10). One patient (patient 13) underwent proximal embolization at the inferior pancreaticoduodenal artery via the superior mesenteric artery to treat pseudoaneurysmal bleeding at the third portion of the duodenum. This case demonstrated the potential advantage of using NBCA as an embolic
agent (Fig 3). Because the microcatheter could not be advanced to the bleeding vessel, the injection was performed proximally. The bleeding artery and the parent artery were both completely occluded. If the parent artery (inferior pancreaticoduodenal artery) on each side of the bleeding vessel had not been embolized, the hemorrhage may have persisted owing to the collateral supply. In this instance, NBCA would likely have been more effective than microcoils unless embolization of different sites could be accomplished. A duodenal mucosal lesion was not demonstrated at panendoscopy in the three patients with duodenal bleeding 9 and 13 days after selective TAE. Acute ischemic insults occurring shortly after TAE may have been missed at the delayed panendoscopic examinations. Although no long-term mucosal complications were encountered in our series, the number of patients with endoscopic evaluation was limited. Further evaluation with more patients is indicated. N-Butyl-2-cyanoacrylate, however, may be a potential alternative to other embolic agents. In conclusion, a small microcatheter can be used with liquid embolic agent to access bleeding vessels. The length and caliber of a bleeding artery, which might cause coils or microcoils to be improperly positioned, do not hamper
216
•
February 2007
Embolization for Acute Upper GI Bleeding using NBCA
embolization with NBCA. When performed by experienced interventional radiologists, the injection of NBCA appears to be a safe alternative to microcoils and embolic particles in the treatment of acute upper gastrointestinal tract bleeding. If the catheters can be placed as close as possible to the site of extravasation, the outcome after NBCA injection is favorable. Even in some cases in which the bleeding artery cannot be catheterized, NBCA is a good alternative to microcoils and embolic particles. In our preliminary experience, TAE with the iodized oil formulation was effective for managing nonvariceal upper gastrointestinal tract bleeding. More cases and experience are needed to verify these findings, and the long-term complications should be monitored. Acknowledgments: This work was supported by a grant from the Department of Medical Research, National Taiwan University Hospital (NTUH095-000382). The
authors acknowledge Dr. Kathy Crabtree for her detailed revision of our article. References 1. Kramer SC, Gorich J, Rilinger N, et al. Embolization for gastrointestinal hemorrhages. Eur Radiol 2000; 10:802– 805. 2. Toyoda H, Nakano S, Kumada T, et al. Estimation of usefulness of N-butyl-2-cyanoacrylate–lipiodol mixture in transcatheter arterial embolization for urgent control of life-threatening massive bleeding from gastric or duodenal ulcer. J Gastroenterol Hepatol 1996; 11: 252–258. 3. Goldman ML, Freeny PC, Tallman JM, et al. Transcatheter vascular occlusion therapy with isobutyl 2-cyanoacrylate (bucrylate) for control of massive upper-gastrointestinal bleeding. Radiology 1978; 129:41– 49. 4. Liu HM, Huang YC, Wang YH. Embolization of cerebral arteriovenous malformations with N-butyl-2-cyanoacrylate. J Formos Med Assoc 2000; 99: 906 –913. 5. Lee KJ, Kim JH, Hahm KB, Cho SW, Park YS. Randomized trial of N-butyl-2-cyanoacrylate compared with in-
6.
7.
8.
9.
10.
JVIR
jection of hypertonic saline-epinephrine in the endoscopic treatment of bleeding peptic ulcers. Endoscopy 2000; 32:505–511. Yok KY, Kum CK, Goh PM. Endoscopic hemostasis of upper gastrointestinal bleeding with histoacryl: last resort before surgery. Endoscopy 1996; 28:256 –258. Kish JW, Katz MD, Marx MV, Harrell DS, Hanks SE. N-butyl cyanoacrylate embolization for control of acute arterial hemorrhage. J Vasc Interv Radiol 2004; 15:689 – 695. Wahl P, Lammer F, Conen D, Schlumpf R, Bock A. Septic complications after injection of N-butyl-2-cyanoacrylate: report of two cases and review. Gastrointest Endosc 2004; 59:911–916. Rosch J, Keller FS, Kozak B, Niles N, Dotter CT. Gelfoam powder embolization of the left gastric artery in treatment of massive small-vessel gastric bleeding. Radiology 1984; 151:365–370. Bandi R, Shetty PC, Sharma RP. Superselective arterial embolization for the treatment of lower gastrointestinal hemorrhage. J Vasc Interv Radiol 2001; 12:1399 –1405.