Carotid artery injury during endonasal surgery is the most feared and catastrophic complication. Internal carotid artery injury is more frequent during skull base surgery, and risk factors include acromegaly, previous revision surgery, and prior radiotherapy and bromocriptine therapy. Nasal packing is frequently used to gain hemostasis, often resulting in vascular occlusion. Recent research recommends the crushed muscle patch treatment as an effect hemostat that maintains vascular patency. Endovascular techniques are recommended for vascular control and complication management. Coil or balloon embolization is preferred in patients with adequate collateral cerebral blood flow, and stent-graft placement or bypass surgery is indicated in those who do not.
| EBM Question | Level of Evidence | Grade of Recommendation |
|---|---|---|
| What factors contribute to ICA injury and what is best management of ICA injury? | 4 | C |
Rupture of the internal carotid artery (ICA) is the most feared and devastating complication of endoscopic sinus and skull base surgery, and may result in death. Injury to the cavernous ICA most commonly results in rupture and overwhelming hemorrhage, with the frequent formation of a pseudoaneurysm. Injury may also cause spasm, thrombosis, embolism, or the formation of a caroticocavernous fistula (CCF) with significant associated morbidity.
Injury to the cavernous ICA is a rare event during endoscopic sinus surgery (ESS). May and colleagues reviewed their experience with ICA injury during ESS and only found 1 case among 4691 patients. Despite the frequency of ESS within the community, a review of the English literature shows a total of only 28 case reports of ICA injury since the advent of the endoscopic approach to the paranasal sinuses ( Table 1 ). The frequency of cavernous ICA injury is much more significant during endonasal, transphenoidal skull base surgery. Raymond and colleagues and Ciric and colleagues showed a 1.1% incidence of ICA injury after the microscopic transphenoidal pituitary approach. More extended endonasal approaches (EEA) center around the management of the internal carotid artery, and not surprisingly have a much higher incidence of ICA injury. Frank and colleagues, Gardner and colleagues, and Couldwell and colleagues reviewed their experience with consecutive EEA resections of craniopharyngiomas, clival chordomas, and chondrosarcomas, showing a 5% to 9% incidence of ICA rupture.
Although experience and knowledge of the relevant anatomy can prevent many potential complications associated with transphenoidal surgery, ICA injury cannot be completely eliminated considering the frequency of these procedures and the increasing complexity of the skull base pathologies encountered. Through review of the endonasal surgical literature, this article focuses on the risk factors for an ICA injury during endonasal surgery, the management of an ICA injury and the complications of this catastrophic surgical event.
Patients at risk
Prevention of the catastrophic bleeding scenario is better than treatment. It is important to recognize the patient that maybe at risk of an ICA injury. The anatomic relationship between the ICA and the sphenoid sinus makes it particularly vulnerable. Fujii and colleagues demonstrated that the bony wall overlying the ICA is not sufficient to protect the artery, at less than 0.5 mm thick. Additionally, in 4 to 22% of cases the lateral sphenoid wall is dehiscent over the carotid with only dura and the sphenoid sinus mucosa separating the ICA from the sphenoid. Renn and Rhoton also found that the ICA bulges into the sphenoid sinus in 71% of cases, and that the artery maybe located as close as 4 mm from the midline. Some authors have found that the distance between the internal carotid arteries within the sphenoid maybe as little as 4 mm, and that the boney sphenoid septum inserts on to the ICA canal wall 16.3% of occasions.
Cavernous ICA anomalies are also not infrequent, with cavernous ICA aneurysm making up 12.8% of all intracranial aneurysms. Some authors have shown an increased incidence of aneurysms in patients with pituitary adenomas, leading some to suggest mechanisms such as mechanical influence, infiltration by the tumor, growth hormone and an IGF-1 effect on the arterial wall. There have been numerous reports of unrecognized pre-operative cavernous ICA aneurysms resulting in ICA rupture. When reviewing all 111 case reports of endonasal cavernous ICA ruptures (see Table 1 ), there are a total of 6 patients that had a pre-operative unrecognized ICA aneurysm. In the 3 patients reported by Koitschev and colleagues, all 3 patients died as a result of uncontrolled hemorrhage, perhaps as a result of a larger defect of the vessel wall with a consecutively higher blood loss.
Numerous authors have linked the association of a number of important patient risk factors associated with a cavernous ICA injury. Raymond and colleagues reviewed their series of 17 ICA injuries showing that 5/17 patients had prior bromocriptine therapy, 5/17 were revision cases, 4/17 had previous radiation therapy and 6/17 pts had acromegaly. Additionally patients with acromegaly tend to have more tortuous and ectatic carotid arteries. While most case reports and series do not discuss the specific case risk factors, a review of the literature (see Table 1 ) demonstrates that it is known that these risk factors contributed in 27 ICA injury cases, some cases with multiple risk factors (revision surgery = 13, radiotherapy = 4, acromegaly = 13, bromocriptine therapy = 4). These features may cause more fibrosis and adherence to the carotid artery, or may simply reflect a more aggressive attempt at complete resection of invasive lesions.
Tumors closely adherent to the ICA require close and careful dissection. Bejjani and colleagues demonstrated that vasospasm occurred in 9 of 470 patients undergoing skull base tumor dissection. In this series vasospasm manifested as altered mental status and/or hemiparesis with risk factors including preoperative embolization, tumor size, vessel encasement/narrowing and total operative time. Three of these patients suffered permanent neurologic deficits. Laws also cautions regarding dissection of tumor away for the cavernous ICA, or displacement of the carotid within the cavernous sinus during attempted hemostasis. They describe 1 fatal, and 2 non-fatal cases as a result of carotid spasm and thrombosis following endonasal transphenoidal surgery.
It is imperative that the ‘at risk’ patient is identified by a thorough pre-operative assessment so that a cavernous ICA injury can be minimized ( Box 1 ). A thorough and careful preoperative assessment of the sella region should be obtained, with the use of a CT scan to help delineate vessel anatomy and its relationship to the sphenoid sinus. MRI scans can demonstrate preoperative ICA aneurysms, with any suspicion confirmed with MRA or digital subtraction angiography.
- •
Anatomic relationships
- ○
Carotid dehiscence
- ○
Sphenoid septal attachment to ICA
- ○
Midline ICA
- ○
- •
Revision surgery
- •
Prior radiotherapy
- •
Prior bromocriptine treatment
- •
Acromegaly
Intra-operative management of a cavernous ICA rupture
Controlling the Surgical Field
Intra-operative ICA rupture creates an immediately challenging surgical field, with a high pressure/high flow bleeding scenario, which may rapidly result in exsanguination of the patient. Massive bleeding leads to a loss of orientation and an obscured surgical field often resulted in the surgeon blindly attempting nasal packing to control the hemorrhage. Additional suction is important to regain orientation of the surgical field. The advantages of the ‘2 surgeon’ skull base team allows for dynamic handling of the endoscope, rather than the single surgeon scenario. Valentine and colleagues have recently developed a reproducible animal model for the carotid artery catastrophe that recreates the intranasal confines of the human nasal cavity, paranasal sinuses and nasal vestibule ( Fig. 1 ). The authors describe their experience with 42 carotid artery injuries and this model, and the surgical steps that enable rapid control the surgical field. The authors relied on the surgical cooperation of both surgeons, acting fluently and quickly to navigate the endoscope’s tip away from the vascular stream and maintaining vision. Frequently a ‘jet of blood’ quickly soiled the endoscopes tip, and the authors found it useful to deliberately place the endoscope into the nostril that afforded some protection offered by the posterior septal edge deflecting the vascular stream into the opposite nostril ( Fig. 2 ). Two large bore suction systems were particularly useful. If the suction instruments were placed below the endoscope (as is routine during ESS) then it frequently results in blood tracking up the suction and soiling the endoscope tip. This frequent occurrence was prevented by placing the suction in the opposite nasal cavity, and allowed the suction to simultaneously guide the vascular stream away from the endoscope tip (in press, laryngoscope).
Intra-Operative Hemostatic Techniques
Every surgical team should have a plan in place should this unexpected complication occur; formulating and executing a plan of action during a crisis is difficult. Emergency surgical ligation has traditionally been used to treat an ICA injury; however, this treatment is often associated with a high incidence of major complications such as death and stroke, and is often an ineffective and harmful treatment. In good collateralization or contralateral compensation the bleeding is likely to still be rapid. Ligation of the internal and external carotid arteries would not only waste time but also block the interventional radiologists’ access to the site of injury.
In the event of unexpected massive bleeding during endonasal surgery then immediate packing is required. A number of techniques have been described and advocated in order the aid this. Some authors advocated for head elevation, and controlled hypotension to reduce the hemorrhage. These measures are likely unnecessary considering the immediate and significant hypotension that will result from massive bleeding while the anesthetic team is trying to implement active resuscitation. If large bore suction devices and the immediate state of hypotension are not enough to keep pace with the bleeding and allow nasal packing then ipsilateral common carotid artery compression is frequently advocated to slow the bleeding rate and can aid the accurate placement of nasal packing. Regarding blood pressure control, Kassam and colleagues, Solares and colleagues and Pepper and colleagues all recommend maintaining normotension through resuscitative measures and fluid replacement to maintain contralateral cerebral perfusion. However, normotension is unlikely to be achieved until the hemorrhage has been controlled. Once vascular control is assured then attention should focus on maintaining adequate cerebral perfusion.
There is a number of different packing agents that have been used during an ICA rupture. A review of the literature demonstrates that gauze packing is overwhelmingly the most frequently used material, likely due to its availability and easy of use. However a number of different agents have been used including Teflon and methyl methacrylate patch, Syvek marine polymer, muscle patch, fibrin glue, gel foam and oxidized cellulose packing, thrombin-gelatin matrix, Oxygel and glue and muslin gauze. Packing materials ideally should be placed with just enough force to control the hemorrhage but not to occlude vascular flow. Absorbable and biocompatible hemostatic agents are advantageous as they don’t require subsequent removal, which can result in re-bleeding if no additional endovascular procedures are undertaken. Raymond and colleagues describe their success with oxidized regenerated cellulose, muscle plugs and tissue adhesives. Profuse intra-operative bleeding occurred in 14 patients and was controlled in all cases, however later reoccurred in 3 patients requiring either a return to theater or endovascular balloon occlusion. Packing was the only method of treatment in 9 patients, with no endovascular treatment, however 1 patient died on day 7 from concurrent basilar artery compression, and another from recurrent tumor at 2 mths of follow-up. The other seven patients had no further bleeding events (follow-up 6mths – 10yrs). Recently Valentine and colleagues compared the hemostatic efficacy of various absorbable and biocompatible materials in the endoscopic carotid artery injury scenario. This study investigated the efficacy of a thrombin-gelatin matrix, oxidized regenerated cellulose and the crushed muscle patch treatment ( Fig. 3 ). Hemostats were applied with just enough force to not compress the artery and allowing ongoing vascular flow. Results demonstrated that the muscle patch treatment achieved rapid hemostasis in all cases, with a mean time to hemostasis of 11 minutes 25 secs. Hemostasis was not achieved in all other topical treatment agents. This evidence strongly supports the use of the crushed muscle patch treatment. However, its application requires careful placement without compression of the vessel, and held in direct contact the vessel defect for approximately 12 minutes. Which packing material to use depends on the size of the vascular defect, what is available within the theater environment and also the past experiences of the surgical team.
Over-packing of the injury site can also be a problem. Endonasal packing often can result in occlusion or stenosis of the cavernous ICA and other major vascular structures. Raymond and colleagues reviewed their angiographic data in 12 patients showing that 8 of 12 had ICA occlusion, and 4 of 12 patients had carotid stenosis. They concluded that over-packing can contribute to the morbidity and mortality of the patient. Laws also concedes that while patency of the ICA is preferred, there our some occasions that the only option is to occlude the ICA with packing and raise the blood pressure in the hope that the collateral circulation will prevent stroke formation.
Direct vascular closure has also been used intra-operatively. Laws described the successful use of direct suture repair in 2 cases, and the use of a sundt-type clip graft, however the details and outcomes of these techniques are not described. Valentine and colleagues recently analyzed the hemostatic efficacy the U-Clip anastomotic device (Medtronic, Jacksonville, FL, USA). This is an endoscopic suturing device that has been successfully used for the suturing of coronary artery vascular anastomosis and for dural reconstructions of the skull base. This device was very effective at achieving hemostasis and maintaining vascular patency however the long-term outcomes remain ( Fig. 4 ).
Unfortunately it is not always possible to achieve intra-operative hemostasis, and transfer to the angiography suite is needed so that endovascular intervention can be performed while the airway is secured. Even though intra-operative hemostasis and vascular control is achieved in most cases, all patients need to have an immediate angiogram so that ICA injury complications can be sought. Angiography should also include the external carotid artery if no abnormality is found within the ICA territory. The otolaryngologist should be available and present to loosen the packing if localization of the ICA injury is not possible due to overtight nasal packing. The optimal management is a balloon test occlusion (BTO); however, this requires a cooperative and awake patient to allow for a full neurologic examination. Awaking the patient and removal of a secure airway is unwise in the face of ongoing ICA bleeding, and hemodynamic instability. Other measures that have been used to determine the presence of adequate collateral flow include analysis of the preoperative MR angiography, transcranial doppler analysis, SPECT imaging and Xenon CT. Even with a well performed and normal BTO there is still a 5–10% risk of delayed infarction after therapeutic carotid artery occlusion.
Endovascular techniques that are available to the interventional radiologist include both balloon and coil embolization, however there are increasing reports of the successful use of endovascular stent-graft placement. Over the last 10yrs transluminal endovascular stent-grafts have increased in popularity, and grafts within the aorta, peripheral vessels and coronary arteries have been reported as safe and effective. Numerous authors recommend the use of endovascular balloon or coil embolization in those patients that have adequate collateral blood flow. Otherwise either an endovascular or surgical bypass procedure is required. Stent-graft placement is advised in those that don’t tolerate ICA occlusion. Some have suggested that all patients have a trial of stent placement, but if this is unsuccessful, then those patients should undergo embolization if tolerated, otherwise a extracranial/intracranial bypass procedure is required.
Intra-operative management of a cavernous ICA rupture
Controlling the Surgical Field
Intra-operative ICA rupture creates an immediately challenging surgical field, with a high pressure/high flow bleeding scenario, which may rapidly result in exsanguination of the patient. Massive bleeding leads to a loss of orientation and an obscured surgical field often resulted in the surgeon blindly attempting nasal packing to control the hemorrhage. Additional suction is important to regain orientation of the surgical field. The advantages of the ‘2 surgeon’ skull base team allows for dynamic handling of the endoscope, rather than the single surgeon scenario. Valentine and colleagues have recently developed a reproducible animal model for the carotid artery catastrophe that recreates the intranasal confines of the human nasal cavity, paranasal sinuses and nasal vestibule ( Fig. 1 ). The authors describe their experience with 42 carotid artery injuries and this model, and the surgical steps that enable rapid control the surgical field. The authors relied on the surgical cooperation of both surgeons, acting fluently and quickly to navigate the endoscope’s tip away from the vascular stream and maintaining vision. Frequently a ‘jet of blood’ quickly soiled the endoscopes tip, and the authors found it useful to deliberately place the endoscope into the nostril that afforded some protection offered by the posterior septal edge deflecting the vascular stream into the opposite nostril ( Fig. 2 ). Two large bore suction systems were particularly useful. If the suction instruments were placed below the endoscope (as is routine during ESS) then it frequently results in blood tracking up the suction and soiling the endoscope tip. This frequent occurrence was prevented by placing the suction in the opposite nasal cavity, and allowed the suction to simultaneously guide the vascular stream away from the endoscope tip (in press, laryngoscope).
Intra-Operative Hemostatic Techniques
Every surgical team should have a plan in place should this unexpected complication occur; formulating and executing a plan of action during a crisis is difficult. Emergency surgical ligation has traditionally been used to treat an ICA injury; however, this treatment is often associated with a high incidence of major complications such as death and stroke, and is often an ineffective and harmful treatment. In good collateralization or contralateral compensation the bleeding is likely to still be rapid. Ligation of the internal and external carotid arteries would not only waste time but also block the interventional radiologists’ access to the site of injury.
In the event of unexpected massive bleeding during endonasal surgery then immediate packing is required. A number of techniques have been described and advocated in order the aid this. Some authors advocated for head elevation, and controlled hypotension to reduce the hemorrhage. These measures are likely unnecessary considering the immediate and significant hypotension that will result from massive bleeding while the anesthetic team is trying to implement active resuscitation. If large bore suction devices and the immediate state of hypotension are not enough to keep pace with the bleeding and allow nasal packing then ipsilateral common carotid artery compression is frequently advocated to slow the bleeding rate and can aid the accurate placement of nasal packing. Regarding blood pressure control, Kassam and colleagues, Solares and colleagues and Pepper and colleagues all recommend maintaining normotension through resuscitative measures and fluid replacement to maintain contralateral cerebral perfusion. However, normotension is unlikely to be achieved until the hemorrhage has been controlled. Once vascular control is assured then attention should focus on maintaining adequate cerebral perfusion.
There is a number of different packing agents that have been used during an ICA rupture. A review of the literature demonstrates that gauze packing is overwhelmingly the most frequently used material, likely due to its availability and easy of use. However a number of different agents have been used including Teflon and methyl methacrylate patch, Syvek marine polymer, muscle patch, fibrin glue, gel foam and oxidized cellulose packing, thrombin-gelatin matrix, Oxygel and glue and muslin gauze. Packing materials ideally should be placed with just enough force to control the hemorrhage but not to occlude vascular flow. Absorbable and biocompatible hemostatic agents are advantageous as they don’t require subsequent removal, which can result in re-bleeding if no additional endovascular procedures are undertaken. Raymond and colleagues describe their success with oxidized regenerated cellulose, muscle plugs and tissue adhesives. Profuse intra-operative bleeding occurred in 14 patients and was controlled in all cases, however later reoccurred in 3 patients requiring either a return to theater or endovascular balloon occlusion. Packing was the only method of treatment in 9 patients, with no endovascular treatment, however 1 patient died on day 7 from concurrent basilar artery compression, and another from recurrent tumor at 2 mths of follow-up. The other seven patients had no further bleeding events (follow-up 6mths – 10yrs). Recently Valentine and colleagues compared the hemostatic efficacy of various absorbable and biocompatible materials in the endoscopic carotid artery injury scenario. This study investigated the efficacy of a thrombin-gelatin matrix, oxidized regenerated cellulose and the crushed muscle patch treatment ( Fig. 3 ). Hemostats were applied with just enough force to not compress the artery and allowing ongoing vascular flow. Results demonstrated that the muscle patch treatment achieved rapid hemostasis in all cases, with a mean time to hemostasis of 11 minutes 25 secs. Hemostasis was not achieved in all other topical treatment agents. This evidence strongly supports the use of the crushed muscle patch treatment. However, its application requires careful placement without compression of the vessel, and held in direct contact the vessel defect for approximately 12 minutes. Which packing material to use depends on the size of the vascular defect, what is available within the theater environment and also the past experiences of the surgical team.
