In all study patients, it was possible to deliver the NBCA mixture to the bleeding site; the angiographic success rate was 100% (32/32). The NBCA mixture was the only embolic material used to achieve angiographic success in 29 patients. Other embolic agents were used in combination with NBCA mixture in three patients, microcoils in two cases and gelatin sponge pieces in one case. Microcoils were used in combination with NBCA for proximal blockade of pseudoaneurysms in two cases during our early experience. Backbleeding from collateral flow after NBCA embolization occurred in one patient with massive gastric cancer bleeding and additional embolization of collateral flow from short gastric artery with gelatin sponge pieces could effectively stop the backbleeding.

Clinical success was achieved in 29 patients (91%). Recurrent bleeding after transcatheter embolization occurred in three patients (9%). However, a second TAE successfully stopped the rebleeding in two patients and emergency surgery was performed in one patient. In two patients with a second TACE, the causes of the bleeding were large duodenal ulcers in both. A repeat angiography revealed that the previously embolized vessels were not recanalized, and that other collateral vessels were responsible for the rebleeding at the same ulcer lesions that were previously embolized.

There was no evidence of symptomatic bowel infarction or other major complications directly related to NBCA embolization in any of the patients. Follow up endoscopic evaluations were performed in 20 patients between 1-105 days (mean 16 days) after the embolization and postembolization bowel ischemia was not observed in any of the patients.

Twenty-three of 32 patients were discharged after clinical improvement without surgery. Two patients underwent surgery for underlying gastric cancer. Seven patients expired during the hospital course due to other underlying conditions. Coagulopathy was present at the time of TAE in 18 patients (56%) and the clinical success rate in this group of patients was 83% (15/18).

Since Rosch et al. (8) introduced the technique of TAE as an alternative to surgery for the control of upper gastrointestinal bleeding in 1972, TAE has gained widespread acceptance for first-line treatment of acute nonvariceal upper gastrointestinal bleeding resistant to endoscopic therapy. TAE is generally preferred over surgery, especially in high-risk patients (3, 6, 9, 10). The efficacy of TAE in initial hemostasis of acute nonvariceal gastrointestinal bleeding has been reported to be 52-90%; however, relatively high rates of rebleeding have also been reported (3, 6, 7, 10-12).

A variety of embolic agents such as gelatin sponge pieces, microcoils, polyvinyl alcohol particles and autologous blood clots have been used in TAE for gastrointestinal bleeding. Some authors have reported the use of liquid adhesives for transcatheter embolization of gastrointestinal bleeding (11, 13-15). However, most interventional radiologists are reluctant to use this technique because of the potential risk of complicated ischemic injury and the difficulty with handling liquid adhesives. Liquid adhesives or glue have been used as embolic agents for nearly three decades, but experience with liquid adhesives outside of neurological interventions has been generally limited (16).

N-Butyl Cyanoacrylate is a well known liquid adhesive used in the vascular system; the US Food and Drug Administration recently approved its use for cerebral arteriovenous malformations in 2000. A localized vascular occlusion mimicking a surgical ligation can be achieved within seconds as the NBCA polymerizes on exposure to the anions in blood (17). Unlike temporary embolic agents, vascular occlusion using NBCA is permanent and cannot be removed by reversal of vessel flow. NBCA is also easy to deliver through a microcatheter because of its low viscosity. However, its lack of radiopacity and its rapid polymerization time make it difficult to use. The addition of iodized oil to NBCA can opacify the agent and polymerization time can be tempered with an increased ratio of iodized oil to NBCA. Stoesslein et al. reported that the mixture of NBCA and iodized oil at a mixing ratio of 1:3-1:4 provided optimal embolic material with a polymerization time of 7.5-11.5 seconds with excellent contrast definition reported in their experimental study in 1982 (18). However, the in vivo polymerization time for NBCA appears to be more rapid than the in vitro time because of the higher body temperature compared to room temperature and a greater presence of anions intravascularly (19). Pollak and white suggested that the estimated in vivo polymerization time for the NBCA to iodized oil mixtures of between 1:1 and 1:4 were 1-4 seconds, with a linear relationship between time and mixture ratio (16). In the present study, we mixed NBCA with iodized oil in ratios varying from 1:1 to 1:3, and we determined the appropriate amount and dilution of NBCA depending on the size and shape of the vessel and its distribution after test injection with contrast material to the target of embolization. However, we used the 1:3 mixing ratio in most cases because the 1:3 ratio was thought to provide a useful polymerization time for embolization in most patients with GI bleeding. The test injection took only one or two minutes and did not cause a significant time delay during emergency procedures.

Among the variety of embolic agents, coils are usually preferred for embolization of pseudoaneurysms. Superselective catheterization and stable catheter positioning are required for safe release and ideal coil embolization. However, sometimes it is difficult to selectively place the catheter close to and beyond the pseudoaneurysm in small or tortuous vessels (20). Furthermore, when there are many efferent arteries originating from the pseudoaneurysm, or many collateral vessels coming from the adjacent arteries, the procedure becomes time consuming and technically difficult to carry out with coils. Although catheterization may be possible, coils may not be delivered because of the tortuosity of vessels (20). However, because of its low viscosity, NBCA can be injected through a microcatheter into small arteries and collateral circulation, which are difficult to embolize with a coil. Even though it is impossible to advance the microcatheter to the bleeding point, when the microcatheter is tightly engaged in the bleeding artery, it is possible to deliver NBCA selectively to the bleeding site without embolizing innocent vessels.

In the upper gastrointestinal tract there are rich anastomoses between adjacent arteries; this had made backbleeding a problem with embolization procedures. For effective hemostasis, embolization of both feeding arteries and potential collateral channels is required. However, gelatin sponge pieces or polyvinyl alcohol particles have a chance to embolize only the feeding artery and sometimes post-embolic angiography reveals backbleeding from collateral channels that are usually more complex and difficult to access (16). Superselection of those collateral feeders is very time consuming and can occasionally result in failed catheterization and embolization. In this series, backbleeding from collateral channels did not occur except in one case of massive gastric cancer bleeding. This suggests that NBCA can effectively embolize all potential collateral channels together with the main feeder vessels with only a single injection. In cases of life threatening bleeding, the choice of NBCA as an embolization material can be lifesaving by minimizing the risk of backbleeding, which can occur with the use of other embolic materials. Toyoda et al. reported that the time for TAE, using NBCA, was significantly faster than for TAE procedures that do not use NBCA. This is important particularly in cases of massive bleeding that require urgent hemostasis (11).

Coagulopathy has been associated with clinical failure after TAE (2, 6, 7). Encarnacion et al. reported that embolization was 2.9 times more likely to fail in patients with coagulopathy, and that death from bleeding after embolization was 9.6 times more likely to occur in this group (6). Schenker et al. also reported that embolization in patients with a coagulopathy at the time of intervention was 2.8 times more likely to fail, and the odds of dying were 3.4 times greater for patients with a coagulopathy (2). In cases of coil or polyvinyl alcohol particle embolization, occlusion of vessels or pseudoaneurysms is dependent on thrombosis rather than the embolic material itself and the coagulation state of the patient is important for the ultimate success of the embolization procedure. However, NBCA does not depend on the coagulation process for its therapeutic effect and therefore can be used effectively in patients with a coagulopathy. Eighteen of 32 patients (56%) in our study had a diagnosis of a coagulopathy prior to arrival to the angiographic suite; the angiographic success rate in this group of patients was 100% and the clinical success rate was 83% in spite of their coagulopathy. Considering that angiographic success rate of TAE in patients with coagulopathy has been previously reported to be less than 46%, NBCA appears to provide a more successful approach in patients with a coagulopathy (2, 6).

Transcatheter embolization with NBCA has the potential risk of complications including: embolization of the proximal segment without adequate embolization at the bleeding site due to rapid polymerization, distal embolization of untargeted sites due to too rapid injection of the NBCA mixture or an excessively prolonged polymerization time (21) and reflux of glue into another innocent vessel by overinjection. Complications related to NBCA embolization can be minimized by careful attention to the specific vascular anatomy and the information obtained from test injections with contrast material. In this series, none of these patients developed symptomatic bowel infarction or mesenteric ischemia after NBCA embolization. These results correlate well with the experience of Yamakado et al. (20) and Kish et al. (13) who similarly had no cases of organ infarction.

In order to carry out a precise and safe embolization and to reduce the possibility of complications with the use of NBCA as an embolic material, we recommend the following: 1) Pay careful attention to the specific vascular anatomy and the information obtained from the test injections with contrast media; 2) Use the wedged catheter technique with no pericatheter flow. It allows the interventional radiologists to precisely control the extent of glue embolization; 3) Perform a simulated injection with contrast media in the same condition before the actual NBCA injection with precise volume calculation.

The single use of a microcatheter, without flushing after NBCA injection, is highly recommended for precise embolization with the wedged technique because overinjection of NBCA can occur during flushing of the microcatheter. The intraluminal volume of microcathter is usually around 0.3 mL and this should be considered for precise volume estimation. For example, if initially 0.8 mL of NBCA mixture was drawn up in a 1-mL syringe and 0.1 mL was left in the syringe after injection, then 0.4 mL was injected into the embolized vessels and 0.3 mL was left in the discarded microcatheter. Adequate experience and training are also mandatory for successful NBCA embolization without major complications. Although most interventional radiologists are familiar with the use of coils and particles such as gelatin sponge pieces and polyvinyl alcohol, few have experience with liquid adhesives such as NBCA (16). NBCA is difficult to use initially, however after becoming familiar with its properties in the animal laboratory, it can be used safely for the care of patients.

In conclusion, TAE with NBCA is a highly effective and safe treatment modality for upper gastrointestinal bleeding, especially in cases when it is technically difficult to advance the microcatheter to the bleeding site and in patients with a coagulopathy.