Secondary Logo

Journal Logo


N-Butyl-2-Cyanoacrylate Pulmonary Embolism After Endoscopic Injection Sclerotherapy for Gastric Variceal Bleeding

Hwang, Seong Su; Kim, Hak Hee; Park, Seog Hee; Kim, Seong Eun; Jung, Jung Im; Ahn, Bo Young; Kim, Sung Hoon; Chung, Soo Kyo; Park, Young Ha; Choi, Kyu Ho

Author Information
Journal of Computer Assisted Tomography: January 2001 - Volume 25 - Issue 1 - p 16-22
  • Free


In patients with portal hypertension, acute gastroesophageal variceal bleeding is one of the main causes of death. Therefore, the treatment and prevention of variceal bleeding are important for successful patient management. For this reason, the demand for palliative treatment of acute gastroesophageal variceal bleeding continually increases. Various treatment modalities such as pharmacological therapy, balloon tamponade, endoscopic injection sclerotherapy (EIS), and endoscopic variceal ligation have all been used for this purpose (1). As n-butyl-2-cyanoacrylate (Histoacryl; Braun-Melsungen, Germany) is rapidly polymerized on contact with blood (2), EIS using n-butyl-2-cyanoacrylate began to be used in the mid-1980s (3,4). A recent report suggested that EIS using n-butyl-2-cyanoacrylate is the initial treatment of choice for gastric variceal bleeding (5). Reported complications after EIS include dysphagia, sepsis, and fever (6). Occasionally, serious complications such as embolization to the brain (7), portal vein (8), and lung (9–12) have been reported. However, the incidence of n-butyl-2-cyanoacrylate pulmonary embolism (PE) following EIS is unclear owing to the small number of reported cases, and the reported severity has varied from asymptomatic (10) to fatal (11). The purpose of our study is to describe the radiologic and clinical manifestations of PE after EIS using n-butyl-2-cyanoacrylate mixed with iodized oil (Lipiodol; Andre Guerbet, Aulnay-sous-Bois, France) for acute gastric variceal bleeding.


Between 1992 and 1999, 140 patients with gastric varices underwent EIS at our institution. There were 114 males and 26 females ranging in age from 7 to 79 years. Variceal bleeding was confirmed by endoscopy. Under fluoroscopic monitoring by the radiologist, direct puncture of the variceal veins and injection of the mixture of n-butyl-2-cyanoacrylate and Lipiodol were performed by the gastroenterologist. At our institution, n-butyl-2-cyanoacrylate was mixed with Lipiodol (1:1) for the purpose of fluoroscopic monitoring and dilution of the n-butyl-2-cyanoacrylate to prevent premature hardening. The n-butyl-2-cyanoacrylate mixed with Lipiodol appears radiopaque on both fluoroscopy and radiography.

The medical records of the 140 study patients were reviewed to evaluate and compare the injected volume of the n-butyl-2-cyanoacrylate mixed with Lipiodol and the presence of respiratory symptoms after EIS in the groups with and without radiopaque pulmonary emboli as seen on chest radiograph or CT. To determine whether high doses of n-butyl-2-cyanoacrylate lead to more PE than low doses, the Student t test was performed between values for the patients with PE and without PE. The sequential changes in respiratory symptoms and the partial arterial oxygen pressure (Pao2) were of particular note in the six patients with PE.

To evaluate the results of EIS, chest and abdominal radiographs were obtained on all patients at admission and immediately after this procedure. The chest radiographs of the 140 patients who underwent EIS were interpreted retrospectively by three radiologists to detect radiopaque pulmonary emboli, and conclusions were reached by consensus. The diagnosis of n-butyl-2-cyanoacrylate PE was established radiographically when radiopaque pulmonary opacities were detected de novo on chest radiography after EIS. Conventional CT scans (Siemens, Erlangen, Germany) were obtained on two patients (Patients 2 and 4), and a high resolution CT scan was obtained on one patient (Patient 1) for detailed evaluation of pulmonary parenchymal changes. In three symptomatic patients with PE (Patients 1, 2, and 3), emergency perfusion scintigraphy (Siemens) with 99mTc-macroaggregated albumin was performed.

Among the six patients with PE, the clinical and chest radiograph follow-up periods ranged from 21 to 53 days. The follow-up chest radiographs, CT scans, and lung perfusion scans were retrospectively evaluated. The shape, distribution, sequential changes of the pulmonary emboli, as well as the PE-associated pulmonary parenchymal changes such as air-space consolidation or mosaic appearance on CT due to different perfusion status were evaluated on the chest radiographs and CT scans. In three patients (Patients 1, 2, and 3), the distribution and size of the perfusion defects on lung perfusion scans were evaluated.


Pre-EIS chest radiographs of the 140 patients who underwent EIS with n-butyl-2-cyanoacrylate mixed with Lipiodol showed no significant findings except for one patient with bilateral pleural effusions. In 6 of the 140 patients (4.3%), multiple newly developed radiopaque pulmonary emboli were detected on immediate post-EIS chest radiographs, and the diagnosis of PE was established. These patients included four men and two women ranging in age from 42 to 60 years.

Comparison Between Groups of Patients With and Without Pulmonary Emboli

The mean volume of the injected n-butyl-2-cyanoacrylate mixed with Lipiodol was 4.16 ± 1.03 ml in patients with pulmonary emboli and 1.76 ± 0.58 ml in patients without. The patients with PE received a significantly higher mean volume of injection (p = 0.0011). Among the six patients with pulmonary emboli, four had respiratory symptoms and the remaining two patients had no respiratory symptoms after the procedure. There were no respiratory symptoms in patients without radiopaque pulmonary emboli after EIS.

Clinical Manifestations in Six Patients with n-Butyl-2-Cyanoacrylate PE

The clinical information on the six patients with n-butyl-2-cyanoacrylate PE is summarized in Table 1. In four of the six patients with PE, respiratory symptoms, including cough, dyspnea, or mild chest discomfort, occurred after EIS. The Pao2 while breathing room air was measured in the four symptomatic patients and ranged from 48 to 86 mm Hg (mean 72.7 mm Hg) at the time of the onset of symptoms. After supportive care with homogeneous oxygen therapy, the patients' respiratory symptoms decreased and Pao2 gradually improved. Patient 4 recovered rapidly (Pao2 = 80 mm Hg) from initial hypoxemia (Pao2 = 48 mm Hg) 1 day following conservative treatment. This patient did not complain of further respiratory symptoms during the remaining 24 days of his hospital stay, but he died owing to hepatic failure from advanced liver cirrhosis.

Clinical summary in six patients with n -butyl-2-cyanoacrylate pulmonary embolism after endoscopic injection sclerotherapy

Chest Radiographic and CT Findings in Patients with n-Butyl-2-cyanoacrylate PE

The radiologic manifestations in the six patients with n-butyl-2-cyanoacrylate PE are summarized in Table 2. On post-EIS chest radiographs and chest CT, the density of the emboli appeared greater than blood but less than calcium. The shape of the emboli was somewhat different from the usual round appearance of pulmonary emboli. Most of the pulmonary emboli were not round but had tubular or nodular radiopacities. Some of the radiopaque emboli were aligned along the pulmonary vessels, and larger emboli were located preferentially in the main and lobar pulmonary arteries (Figs. 1A and 2A). The size and number of the pulmonary emboli varied. Roughly measured, there were >20 emboli in each of two patients and <20 emboli in each of the four remaining patients.

Radiological summary in six patients with n -butyl-2-cyanoacrylate pulmonary embolism after endoscopic injection sclerotherapy (EIS)
FIG. 1.
FIG. 1.:
Patient 1. A 42-year-old woman with pulmonary embolism after endoscopic injection sclerotherapy (EIS). A: Chest radiograph obtained immediately after EIS shows multiple tubular and nodular radiopacities in both parahilar regions. B: Chest CT scan obtained 2 days after EIS reveals multiple tubular and nodular radiopacities within the pulmonary arteries (arrows). Note no significant parenchymal changes associated with radiopaque pulmonary emboli. C: 99mTc-Macroaggregated albumin lung perfusion scan obtained 3 days after EIS reveals multiple wedge-shaped, subsegmental photon defects at both the upper and the right midlung zones (arrows). D: Follow-up chest radiograph 21 days after EIS shows that the initially noted pulmonary radiopacities are markedly decreased in both size and number (arrows).
Figure 1
Figure 1:
Figure 1
Figure 1:
Figure 1
Figure 1:
FIG. 2.
FIG. 2.:
Patient 4. A 60-year-old man with pulmonary embolism after endoscopic injection sclerotherapy (EIS). A: Chest radiograph obtained immediately following EIS reveals the totally opacified left lingular segmental artery and multiple nodular radiopacities in both lower lung regions. Also noted are preexisting bilateral pleural effusions and an area of patchy increased density in the right midlung zone, representing a collapsed right middle lobe before EIS. B and C: Chest CT scan with mediastinal (B) and lung window (C) obtained 1 day after EIS demonstrates no evidence of pulmonary parenchymal changes associated with pulmonary emboli. Preexisting bilateral pleural effusions are again noted.
Figure 2
Figure 2:
Figure 2
Figure 2:

Chest radiographs and CT scans indicated the absence of subtle pulmonary parenchymal changes such as infarctions or mosaic attenuation caused by pulmonary emboli (Figs. 1B and 2B and C). A comparison of chest radiography and CT for detecting pulmonary emboli was not included in this study because chest radiography obviously reveals radiopaque emboli, and the acquisition protocol for CT scanning was intended to detect subtle pulmonary parenchymal changes, not pulmonary emboli.

The collateral venous route responsible for PE was visualized on CT and abdominal radiography in two patients (Patients 2 and 4). The n-butyl-2-cyanoacrylate mixed with Lipiodol embolized to the lung via the gastrorenal and splenorenal shunts-inferior vena cava pathway (Fig. 3).

FIG. 3.
FIG. 3.:
Patient 2. A 48-year-old man with pulmonary embolism after endoscopic injection sclerotherapy (EIS). A: Abdominal radiograph obtained immediately following EIS shows opacified gastric varices draining into the inferior vena cava (black arrows) via the gastrorenal-splenorenal shunt (white arrows). IVC, inferior vena cava; rv, left renal vein; grs, gastrorenal shunt. B and C: Abdominal CT scan obtained 1 day after EIS demonstrates opacified inferior vena cava (thin black arrow in B and white arrow in C), gastric varices (thick black arrow in B), and gastrorenal shunt (thick white arrow in C).
Figure 3
Figure 3:
Figure 3
Figure 3:

Perfusion Scintigraphic Findings in Three Patients with n-Butyl-2-cyanoacrylate PE

Perfusion scintigraphy (Patients 1, 2, and 3) showed multiple wedge-shaped, peripheral, subsegmental photon defects (Fig. 1C). There were no lobar or segmental perfusion defects, especially in the one instance of total occlusion of the left lingular segmental artery (Patient 4). Correlated with the number of pulmonary emboli detected on chest radiography, the number of perfusion defects tended to be greater in patients with a large number of emboli. All perfusion defects were wedge-shaped, subsegmental defects.

Chest Radiographic Follow-up in Six Patients with n-Butyl-2-cyanoacrylate PE

The multiple radiopaque pulmonary emboli in five of the six patients with PE gradually decreased in number, size, and attenuation by the time of the 1 month follow-up examination (Fig. 1D). Interestingly, although the patient's respiratory symptoms no longer existed, follow-up chest radiographs revealed little interval change in the size, number, or attenuation of the pulmonary emboli during the intervening 21 days (Patient 6).


EIS with n-butyl-2-cyanoacrylate is being increasingly used for emergent control of acute gastric variceal bleeding, whereas endoscopic variceal ligation or injection sclerotherapy is widely used to control active esophageal variceal bleeding. However, in cases of gastric variceal bleeding, it is difficult to control the bleeding because of the somewhat different hemodynamic status, such as the presence of rich collateral channels and a greater amount of blood flow (1,13). In addition, a 5:5 or 7:3 dilution of radiolucent n-butyl-2-cyanoacrylate with radiopaque Lipiodol has achieved better radiographic opacity and has delayed polymerization time during EIS (2,5). Therefore, the risk of accidental PE during EIS for gastric varices almost always exists and may, in fact, be increasing owing to the greater demand for sclerotherapy. Embolized n-butyl-2-cyanoacrylate causes a harmful reaction in the host tissue. Previous reports have described this response, which induces acute and chronic inflammatory changes, including perivascular inflammation and vessel wall necrosis with a chronic foreign body reaction (14,15). Embolized Lipiodol may also produce lung damage such as that caused by chemical injury, resulting from the breakdown of the embolized Lipiodol into free fatty acids (16).

In a series of 100 patients with a variety of lesions embolized with isobutyl-2-cyanoacrylate, two patients were clinically asymptomatic whereas one patient who underwent isobutyl-2-cyanoacrylate/ethiodized oil embolization died of respiratory distress (17). A previous report on iodized oil PE described six patients who developed respiratory symptoms and one patient who died of respiratory arrest among 336 patients in whom transarterial chemoembolization of hepatocellular carcinoma was undertaken (16). In our series of 140 patients with EIS in whom n-butyl-2-cyanoacrylate mixed with Lipiodol was used, 6 patients had PE, but 2 of these were asymptomatic. The initial symptoms may have been caused by mechanical obstruction of the pulmonary arteries with an acute inflammatory reaction to the n-butyl-2-cyanoacrylate and may also have been aggravated by the chemotoxic effects of the Lipiodol (18). We believe that in our study, the incidence of actually damaged lung caused by pulmonary emboli was relatively small because of the absence of a fatal respiratory disturbance clinically. It is possible that pulmonary collateral perfusions near the occluded areas may have decreased the patients' symptoms as well as the perfusion defects on scintigraphy.

Although the amount of n-butyl-2-cyanoacrylate mixed with Lipiodol generally used during EIS is <3 ml to prevent occlusion of undesired vessels or inadvertent embolism, the actual amount varies from case to case (2,5). To our knowledge, no study has been reported regarding the difference of injected volume of n-butyl-2-cyanoacrylate between patients with PE and without PE after EIS, presumably because of the unusual occurrence of PE after EIS. Despite the relatively small number of PEs in our study, a significantly higher volume of n-butyl-2-cyanoacrylate mixture had been injected into the patients with PE, that is, 4.2 vs. 1.8 ml in the patients without PE (p = 0.0011). Therefore, we believe that when the injected volume of n-butyl-2-cyanoacrylate mixture is increased, the potential risk of PE may also increase.

Chest radiography and CT scans showed varying sizes of pulmonary arterial emboli represented by either tubular or nodular radiopacities. Despite the absence of a predominant location of the pulmonary emboli on chest radiographs, the larger emboli were invariably located in the pulmonary hilar region. This may reveal preferential involvement of the main or lobar pulmonary arteries probably causing few perfusion defects at pulmonary scintigraphy. Furthermore, lung perfusion scans with peripheral and subsegmental perfusion defects suggest that the pulmonary perfusion defects were probably caused by tiny distal pulmonary emboli rather than by larger emboli adjacent to the pulmonary hilum. These tiny pulmonary emboli may be subtle or not visualized on chest radiography or CT scans owing to their overlap with underlying pulmonary vascular shadows on chest radiography or to volume averaging in our CT slice thickness and interval. Thus, it is possible that there are asymptomatic PE patients who have radiographically undetected microemboli. Despite the possibility of undetected or missed pulmonary emboli on chest radiography, we believe that most of these microemboli may cause subclinical PE.

The typical collateral venous pathway in portal hypertension is illustrated in Fig. 4. Portosystemic anastomoses frequently occur via the umbilical and paraumbilical veins, the gastroesophageal veins, the gastrorenal-splenorenal veins, or the collateral veins through the superior or inferior mesenteric vein (19). During EIS for gastric varices, an excessive amount of n-butyl-2-cyanoacrylate mixed with Lipiodol may be embolized into the systemic circulation via collateral venous channels such as the gastrorenal-splenorenal veins, which are directly connected to the gastric varices. In Patients 2 and 4, the gastrorenal-splenorenal shunt and the inferior vena cava were opacified during and after EIS. Thus, during EIS, the opacification of unwanted collateral vessels may indicate not only the possible occlusion of unwanted vessels but also the risk of possible PE. Recently, a case of pulmonary, cerebral, and coronary embolism during EIS was reported, and the patient had a patent foramen ovale (12). Despite the absence of systemically embolized patients in our cases, the possibility of systemic embolization during EIS via the patent foramen ovale should be considered as it occurs in 18% of the general population during routine echocardiography (20).

FIG. 4.
FIG. 4.:
Common collateral venous systems in portal hypertension. In portal hypertension, commonly occurring collateral venous channels include the umbilical and paraumbilical veins (UV), the gastroesophageal veins (GV and EV), the gastrorenal-splenorenal veins (GRSRV), and the collateral veins through the superior or inferior mesenteric vein (SMV or IMV). GV, gastric veins; EV, esophageal veins; UV, umbilical and paraumbilical veins; GRSRV, gastrorenal-splenorenal veins; HV, hemorrhoidal veins; PV, portal vein; SV, splenic vein; IMV, inferior mesenteric vein; SMV, superior mesenteric vein; IVC, inferior vena cava.

In conclusion, radiographically evident n-butyl-2-cyanoacrylate PEs are uncommonly observed following EIS and are represented by tubular or nodular radiopacities on chest radiography or CT. PEs appear to be more common in patients receiving a higher volume of liquid acrylate during the procedure. Affected patients were either mildly symptomatic or asymptomatic, and there were no direct fatalities of this complication.


The authors thank In Hye Park, Ph.D. (Department of Biostatistics, Catholic University of Korea, Seoul, Korea), for her statistical assistance and Bonnie Hami, M.A. (Department of Radiology, University Hospitals of Cleveland, Cleveland, OH, U.S.A.), for her editorial assistance in the preparation of this manuscript.


1. D'Amico G, Pagliaro L, Bosch J. The treatment of portal hypertension: a meta-analytic review. Hepatology 1995; 22:332–54.
2. Soehendra N, Grimm H, Nam VCh, et al. n-Butyl-2-cyanoacrylate: a supplement to endoscopic injection therapy. Endoscopy 1987; 19:221–4.
3. Gotlib JP, Zimmermann P. Une nouvelle technique de traitement endoscopique des varices oesophagienes; l'obliteration. Endosc Dig 1984; 7:10–2.
4. Soehendra N, Nam V, Grimm H, et al. Endoscopic obliteration of large esophagogastric varices with bucrylate. Endoscopy 1986; 19:221–4.
5. Ogawa K, Ishkawa S, Narita Y, et al. Clinical evaluation of endoscopic injection sclerotherapy using n-butyl-2-cyanoacrylate for gastric variceal bleeding. J Gastroenterol Hepatol 1999; 14:245–50.
6. Gotlib JP. Endoscopic obturation of esophageal and gastric varices with a cyanoacrylic tissue adhesive. Can J Gastroenterol 1990; 9:637–8.
7. See A, Florent C, Lamy F, et al. Cerebrovascular accidents after endoscopic obliteration of oesphageal varices with isobutyl-2-cyanoacylate in 2 patients. Gastroenterol Clin Biol 1986; 10:604–7.
8. Moustafa I, Omar MM, Nooh A. Endoscopic control of gastric variceal bleeding with butyl cyanoacrylate. Endoscopy 1993; 25:A11.
9. Thakeb F, Kader S, Salama Z, et al. The value of the combined use of n-butyl-2-cyanoacrylate and ethanolamine oleate in the management of bleeding esophagogastric varices. Endoscopy 1993; 25:A5.
10. Takasugi JE, Shaw C. Inadvertent bucrylate pulmonary embolization: a case report. J Thorac Imag 1989; 4:71–3.
11. Moustafa I, Omar MM, Nouh A. Endoscopic control of gastric variceal bleeding with butyl cyanoacrylate in patients with schistosomiasis. J Egypt Soc Parasitol 1997; 27:405–10.
12. Roesh W, Rexroth G. Pulmonary, cerebral and coronary emboli during bucrylate injection of bleeding fundic varices. Endoscopy 1998; 30:S89–90.
13. Watanabe K, Kimura K, Matsutari E, et al. Portal hemodynamics in patients with gastric varices. A study in 230 patients with esophageal and/or gastric varices using portal vein catheterization. Gastroenterology 1988; 95:434–40.
14. Cromwell LD, Freeny PC, Kerber CW, et al. Histologic analysis of tissue response to bucrylate-pantopaque mixture. AJR 1986; 147:627–31.
15. Brothers MF, Kaufmann JCW, Fox AJ, et al. n-Butyl-2-cyanoacrylate-substitute for IBCA in interventional neuroradiology: histopathologic and polymerization time studies. AJNR 1989; 10:777–86.
16. Chung JW, Park JH, Im JG, et al. Pulmonary oil embolism after transcatheter oily chemo-embolization of hepatocellular carcinoma. Radiology 1993; 187:689–93.
17. Goldman ML, Philip PK, Sarrafizadeh MS, et al. Transcatheter embolization with bucrylate (in 100 patients). Radiographics 1982; 2:340–75.
18. Pelz DM, Lownie SP, Fox AJ, et al. Symptomatic pulmonary complications from liquid acrylate embolization of brain arteriovenous malformations. AJNR 1995; 16:19–26.
19. Madrazo B, Jafri SZ, Shirkhoda A, et al. Portosystemic collaterals: evaluation with color Doppler imaging and correlation with CT and MRI. Semin Intervent Radiol 1990; 7:169–84.
20. Brodhun R, Volles E. Paradoxe zerebrale embolie bei offenem foramen ovale. Notfallmedizin 1998; 24:20–4.

n-Butyl-2-cyanoacrylate; Lungs, embolism; Gastrointestinal tract, bleeding; Endoscopy; Computed tomography

© 2001 Lippincott Williams & Wilkins, Inc.