Early diagnosis and adequate management of post-tonsillectomy hemorrhage is mandatory to prevent hemorrhagic shock and exsanguination . Treatment options include local maneuvers, surgical ligation, and endovascular embolization. Endovascular embolization has some advantages. First, the initial diagnostic angiogram can be performed to detect the bleeding vessel followed by superselective occlusion of the injured artery in the same session. Second, it avoids the risk of injury of the vagal and accessory nerves . In our study, the patients were sent to our unit after failure of the local maneuvers. No one of the patients underwent further surgery.
Van Cruijsen et al.  have stated that post-tonsillectomy pseudoaneurysm only involves children under the age of 10 years, due to the smaller anatomy and thinner pharyngeal muscles. However, in the current study, there was an adult patient. Studies by Pourhassan et al.  and Walshe et al.  have reported that the possibility of post-tonsillectomy pseudoaneurysm formation is not restricted by age.
The ascending pharyngeal artery, dorsal lingual artery, inferior tonsillar branch of facial artery, superior tonsillar branch of the descending palatine artery, and the ascending palatine artery supply the tonsils and are all at risk to develop a pseudoaneurysm . In our study, the site of injury was facial artery in four patients, lingual artery in three patients, and linguofacial trunk in one patient. Most of the previous studies reported that lingual artery was most common site of injury in their studies [3, 4, 13]. To our knowledge, there are few published cases of facial artery injury as source of bleeding. Juszkat et al.  and Choi et al.  reported cases of recurrent post-tonsillectomy bleeding due to an iatrogenic facial artery pseudoaneurysm.
In our study, the post-operative onset of bleeding was from 4 to 10 days postoperatively. This corresponds with the study of Manzato et al.  that reported that cases of hemorrhage from pseudoaneurysm have presented in the range of 1–3 weeks and Hassan et al.  who stated that the bleeding started from 3 to 11 days after surgery. Moreover, there are some cases present over a month after surgery and so the possibility of pseudoaneurysm should remain even beyond the typical window of post-tonsillectomy hemorrhage .
In the current study, the eight patients were treated by endovascular embolization using concentrated NBCA glue diluted with Lipiodol. This matched with study done by Hassan et al. . Manzato et al.  recommended trapping of the injured segment by microcoils to avoid possible distal migration of fluid or particulate embolic materials, such as glue or PVA particles. By using NBCA glue, we can achieve permanent occlusive effect, with a decrease possibility of recanalization, in contrast to possible delayed coil extrusion which may occur with coils [3, 16]. Proximal occlusion of the injured artery by coils may lead to retrograde filling from collateral pathways resulting in recurrent bleeding; so, trapping of the injured segment should be done [7, 8]. For trapping of the pseudoaneurysm by microcoils, we should bypass the site of injury by the microcatheter. This will increase the risk of rupture of the pseudoaneurysm, and sometimes it could be technically difficult. The use of glue allows injection with the microcatheter situated just proximal to the pseudoaneurysm or slightly within it, infiltrating the whole segment without the need of pushing the microcatheter distally . The glue has low cost as compared to microcoils and onyx. The histoacryl glue (NBCA) has a faster rate of polymerization than other liquid embolic agents as Onyx, so decreasing the possibility of distal migration and occlusion of the normal vessels .
Possible major complications of using histoacryl glue (NBCA) include inadvertent distal migration into the normal branches of the artery harboring the pseudoaneurysm. Also, proximal reflux into the main ECA or its other branches could occur. This can lead to local ischemic changes as tongue or lip necrosis . In our study, we used concentrated histoacryl glue (NBCA) to avoid distal extension and non-target embolization. Another major possible complication is the extension of NBCA to the internal carotid artery either through retrograde reflux or through potentially dangerous anastomoses between the external carotid artery and internal carotid artery which may lead to cerebrovascular accidents and blindness. The possible existence of anastomoses determines the type of the embolizing agent that should be used [6, 7]. In all cases, we assess the pre embolization angiograms for any possible anastomosis with intracranial circulation. In our study, there was no major complication. In one patient with pseudoaneurysm at the ostium of the facial artery, inadvertently proximal reflux into the adjacent part of the ECA occurred leading to its occlusion, yet the patient did not experience any immediate or delayed complication due to the rich anastomosis and collaterals at this area. In the study done by Hassan et al.  in one patient, active bleeding occurred during the diagnostic angiogram followed by rapid injection of the glue leading to distal extension of glue into distal branches yet with no related complications.
The limitations of our study included the small number of patients and being a retrospective study.