Clinical manifestations

Presentation and course

The clinical manifestations of a carotid-cavernous fistula can be predicted by the structures within the cavernous sinus that are affected by the fistula. Therefore, a carotid-cavernous fistula may present with palsies of cranial nerves III, IV, V1, V2, or VI. In addition, anterograde venous flow may cause orbital and ocular venous congestion, resulting in proptosis, chemosis, arterialized venous loops on the sclera, increased intraocular pressure with subsequent glaucomatous vision loss, and engorgement of the extraocular muscles resulting in further restriction of eye movement.

Direct fistulas present dramatically due to the large volume of blood that is being shunted.

Although fistula occurs most commonly on the side of head trauma, it may rarely occur on the contralateral side (29). In a series of 98 consecutive patients presenting with direct fistulas, the most common symptom was a prominent orbital bruit (80%), followed by proptosis (72%), and chemosis (55%). Isolated ipsilateral abducens palsy occurred in 49%, retro-orbital pain or headaches in 24%, ophthalmoplegia in 23%, and vision loss in 17% (52).

Intracranial hemorrhage may develop in 5% of patients with direct carotid-cavernous fistula. About 2% of these types of intracranial hemorrhages can present with life-threatening nosebleeds (91).

A rare presentation of direct fistula is brainstem venous congestion, resulting in reduced consciousness, facial numbness, dysarthria, and gait ataxia (14). Unilateral limb weakness was observed in a patient with contralateral basal ganglia congestion (31).

By contrast, indirect (“dural”) carotid-cavernous fistulas have a more subtle presentation due to the smaller volume of shunted blood. These patients are typically middle-aged or elderly women. Posteriorly draining indirect carotid-cavernous fistulas lack the proptosis or the orbital bruit. The lack of conjunctival vessel arterialization gives the indirect carotid-cavernous fistula the moniker “white-eyed shunt” (01). These fistulas are often asymptomatic and may present with sudden, isolated ocular motor nerve palsy and are almost always associated with ocular or orbital pain. The third nerve is the most involved cranial nerve and can be complete with pupil involvement, incomplete with pupil involvement, or incomplete with pupil sparing.

This presentation of isolated, pupil-involving third nerve palsy without orbital congestion may be confused with a cerebral aneurysm or microvascular palsy. However, the true diagnosis of a posterior-draining indirect carotid-cavernous fistula will be made after cerebral angiography.

In contrast, anterior-draining indirect fistulas usually produce orbital congestion and visual symptoms and manifest similarly to direct carotid-cavernous fistulas. This manifests as arterialization of the conjunctival vessels, proptosis, chemosis, ocular motor paresis, and orbital pain. Oftentimes conjunctival injection is mistaken for “chronic conjunctivitis.” However, close inspection reveals the characteristic radial pattern of arterialization seen in carotid-cavernous fistula.

Increased intraocular pressure, retinal vessel occlusion, ischemic optic neuropathy, or proliferative retinopathy lead to tractional retinal detachment and vision loss in 30% of patients (61). Increased intraocular pressure also can be a result of angle closure from massive choroidal effusion (88). However, not all anteriorly draining indirect carotid-cavernous fistulas present with ocular or orbital signs. A woman with anteriorly draining indirect carotid-cavernous fistula presented only with headache and isolated abducens nerve palsy. In this case, the anastomosis between the superior ophthalmic vein and angular facial vein prevented pressure buildup and the typical ocular or orbital signs (42).

Another distinguishing feature of indirect carotid-cavernous fistula is the waxing and waning nature. Up to 60% of indirect carotid-cavernous fistulas close or improve spontaneously (79). Spontaneous closure of an indirect fistula site may even occur in patients who present with florid clinical symptoms such as optic neuropathy and orbital congestion (13). Improvement is thought to be due to spontaneous thrombosis of the fistula site. Recurrences or changes in clinical symptoms may occur as these clots dissolve or migrate, thus, diverting blood flow to different areas.

A retrospective study of 47 patients with cavernous-carotid fistulas (26 indirect, 21 direct) revealed a similar presentation of headache and ophthalmic symptoms and signs (proptosis, conjunctival injection, ophthalmoparesis, chemosis, visual acuity loss) between the 2 groups (33). The average age of patients with indirect fistulae was 62 years, compared to 48 years for those with direct fistulas. Indirect fistulas were more likely associated with elevated intraocular pressure and fundus abnormalities, whereas direct fistulas with a bruit and history of trauma. Patients who presented with vision loss were 3 times more likely to have residual symptoms, independent of carotid-cavernous fistula classification.

Prognosis and complications

The overall prognosis of a carotid-cavernous is fistula good, with over 80% experiencing complete cure after endovascular treatment (28). The main cause of death with indirect carotid-cavernous fistulas is intracerebral hemorrhage, which occurs at a rate of approximately 1.8% per year (22). Up to 60% of indirect carotid-cavernous fistulas will close spontaneously or after diagnostic angiography alone, thus, obviating the need for intervention (79; 15). Close observation of indirect carotid-cavernous fistula for as long as the ocular symptoms remain mild and there are no neurologic symptoms is reasonable. Elevated intraocular pressure that occurs in indirect fistulas may respond to medical management alone. In 1 series of 14 eyes, 9 had elevated intraocular pressure of which only 3 had unfavorable intraocular pressure control. Only 1 of these eyes required glaucoma surgery (43). However, vision loss can have other causes like irreversible ischemic optic neuropathy, proliferative retinopathy, and choroidal detachment and effusion (37). If the patient develops uncontrollably high intraocular pressure, intractable exposure keratopathy from proptosis, unacceptable diplopia, or intracerebral hemorrhage, the fistula should be closed.

Direct fistulas have higher morbidity and mortality compared to indirect fistulas. Out of 127 patients with direct fistulas, 3.9% died, whereas no one died from indirect fistulas. Additionally, the rates of intracerebral or subarachnoid hemorrhage (3.1% vs. 0%), life-threatening epistaxis (3.1% vs. 0%), and increased intracranial pressure (8.7% vs. 3.6%) were much higher for direct versus indirect carotid-cavernous fistulas. Subarachnoid hemorrhage from a direct carotid-cavernous fistula portends a particularly grim prognosis, with all 4 of these patients dying quickly (35).

Clinical vignette

History. A 47-year-old man was hit by a car several months before evaluation. Since the accident, he had developed swelling and redness in the left eye. The eye hurt when exposed to light.

Physical examination. The physical examination showed visual acuity of 20/40 bilaterally. He had gross proptosis and periorbital swelling of the left eye, and a thrill was present over the left eye. The intraocular pressure was 8 mm Hg in the right eye and 16 mm Hg in the left eye.

Diagnostic tests (angiography). A catheter angiogram revealed a direct carotid-cavernous sinus fistula located at the left proximal posterior genu and cavernous sinus. There was significant dilatation of the superior ophthalmic vein. There was so much shunting that there was little to no filling in the left hemispheric arteries. This was diagnosed as a type A carotid-cavernous sinus fistula.

Treatment. The fistula was treated with a microcatheter that was guided up the internal carotid artery and through the rent in the carotid to the superior ophthalmic vein. Multiple coils were placed in the superior ophthalmic vein. As the coils filled the superior ophthalmic vein, shunting decreased and the arteries of the left hemisphere became visible. Once the ophthalmic vein was coiled, a stent was placed in the internal carotid artery over the rent to prevent coils from falling back into the parent vessel. Additional coils were then placed to occlude the cavernous sinus.

Outcome. The patient’s pain proptosis and chemosis significantly improved after the procedure. He had no restriction of extraocular movements, and the diplopia resolved.

Biological basis

Etiology and pathogenesis

Direct carotid-cavernous fistulas are usually associated with head trauma, surgery, or endovascular intervention. Direct fistulas were reported after stent-assisted coil embolization of an intracavernous carotid aneurysm, mechanical thrombectomy for stroke, and after craniofacial surgery (15; 02; 98).

The mechanism of direct carotid-cavernous fistula formation is uncertain. Some authors believe that these fistulas result from bony fractures tearing the intracavernous internal carotid artery from its dural attachment or from shearing of this vessel due to inertia (74). Others believe that direct carotid-cavernous fistulas are the result of increased intraluminal pressure resulting from abrupt compression of the carotid arteries (eg, during sudden neck flexion). Autopsy did not reveal fractures of the bony walls surrounding the intracavernous internal carotid artery. Rupture of the vessel occurred most often in the C4 segment of the internal carotid artery (27 of 42 patients) followed by the C2 (13 of 42 patients) and C3 (2 of 42 patients) segments. Because the C4 segment is free of the trabeculae of the cavernous sinus, Parkinson’s “tearing” hypothesis cannot completely account for fistula formation (39).

Besides trauma, direct carotid-cavernous fistulas can occur spontaneously in conditions that weaken the internal carotid artery wall, such as aneurysm, Ehlers-Danlos syndrome (63), fibromuscular dysplasia (41), pseudoxanthoma elasticum (80), or osteogenesis imperfecta (26).

Indirect carotid-cavernous fistulas usually occur spontaneously and insidiously in middle-aged to elderly women. It is believed that indirect fistulas arise from rupture of the thin-walled dural arteries that normally cross the cavernous sinus (69). The currently favored theory is that these fistulas represent collateral anastomoses occurring after spontaneous venous thrombosis of vessels in the cavernous sinus in patients with a thrombotic tendency (89; 81). This may explain why dural fistulas are more common in women especially during pregnancy and the early postpartum period (93). Additionally, a congenital defect is suggested by indirect fistulas in infants without an obvious risk factor (51; 30). Unlike direct fistulas, indirect fistulas rarely follows with trauma.

Epidemiology

Direct carotid-cavernous fistulas occur with an incidence ranging from 0.17% to 1.01% after traumatic brain and facial injury (87). In a series of 100 consecutive direct carotid-cavernous fistulas, 76 were caused by trauma, 22 by a ruptured aneurysm, and 2 were iatrogenic. Motor vehicle accidents (54 cases) and falls (10 cases) were the most common causes of trauma in this series. Most trauma-related direct carotid-cavernous fistulas occurred in young men (57%; mean age of 36 years), whereas the ruptured aneurysm-related fistulas occurred predominantly in older women (77%; mean age of 54 years) (52).

A retrospective chart review of 312 patients showed that in basal skull fractures, the incidence of direct carotid-cavernous fistula is 3.8% and up to 8.3%; the middle cranial fossa is involved (53). In another retrospective study, 11.5% of patients developed a direct cavernous carotid fistula within 2 weeks of treatment of a cavernous aneurysm with a flow diverter (82).

Intracranial dural arteriovenous fistulas account for 10% to 15% of all intracranial arteriovenous malformations (68). Of these, indirect carotid-cavernous fistulas account for approximately 12% and are the second most common type of intracranial dural fistula after those involving the transverse sigmoid sinus. One study suggests that 0.1% of the United States population may have an arteriovenous malformation (12). Thus, at most 0.0018% (approximately 5400 people) of the United States population would be expected to have an indirect carotid-cavernous fistula.

Prevention

Most direct carotid-cavernous fistulas occur after trauma. Therefore, protection against head trauma is reasonable. A high level of suspicion for a fistula is warranted when severe head trauma is associated with typical neuro-ophthalmological signs. Iatrogenic related direct fistula can be prevented best by thorough knowledge of the patient’s carotid vasculature and medical history. For example, inadvertent creation of a direct fistula during transsphenoidal surgery for pituitary adenoma is more common in patients with a history of previous transsphenoidal surgery or radiation treatment. Assessing patient’s unique vasculature with magnetic resonance or digital subtraction angiography before tumor resection may inadvertent creation of a direct carotid-cavernous fistula during operation (50).

Although there is no proven way to reduce the incidence of indirect carotid-cavernous fistulas, treating hypertension, atherosclerosis, and thrombophilia seems sensible.

Differential diagnosis

Because direct carotid-cavernous fistulas typically present with prominent orbital or ocular congestion, the main differential diagnoses include orbital processes such as thyroid eye (Graves) disease, idiopathic orbital inflammation (inflammatory orbital pseudotumor), orbital cellulitis, intraorbital tumor (eg, meningioma), or the various other causes of cavernous sinus syndrome (eg, carotid-cavernous thrombosis or Tolosa-Hunt syndrome). The main factor that distinguishes direct carotid-cavernous fistulas from these other entities is its typical association with acute, severe head trauma.

In contrast, indirect carotid-cavernous fistulas present more insidiously, and their clinical manifestations may be overlooked or confused with benign entities such as conjunctivitis. One specific sign that distinguishes both direct and indirect carotid-cavernous fistulas from other disease entities is the presence of a characteristic orbital bruit. In almost all cases, the appropriate imaging study will reveal the diagnosis. In absence of imaging, the following pearls can help distinguish carotid-cavernous fistulas from other masqueraders:

Direct carotid-cavernous fistulas.

Thyroid eye disease. Though thyroid eye disease may present with exophthalmos and limitation of ocular movement, it usually does so in a subacute manner and without trauma. In addition, it typically involves both eyes (although asymmetrically) with upper eyelid retraction being its hallmark. Chemosis and conjunctival injection are less common and not as prominent compared to direct carotid-cavernous fistulas (11).

Idiopathic orbital inflammation. This entity encompasses all nonspecific orbital inflammation that cannot be attributed to infection, tumor, or another identifiable cause. Signs and symptoms depend on which orbital structures are involved, ranging from a mild myositis to a severe orbital apex syndrome. It also may present acutely, subacutely, or chronically and, thus, may mimic both direct and indirect carotid-cavernous fistulas. Like a direct carotid-cavernous fistula, it usually presents unilaterally and can have severe orbital signs as well as severe scleral injection from scleritis. However, in contrast to a direct carotid-cavernous fistula, it is not typically associated with trauma or an orbital bruit. Idiopathic orbital inflammation is also exceptionally sensitive to corticosteroids unlike most of the entities listed in this differential (44).

Orbital cellulitis. This condition may also present with severe proptosis, eyelid edema, pain with extraocular eye movements, and vision loss. However, infection occurs over several days and often have a history of upper respiratory infection, nasal congestion, or sinusitis preceding the orbital signs. In addition, these patients typically have constitutional symptoms such as fever, decrease in appetite, or malaise. This entity typically will respond to the appropriate intravenous antibiotic (92).

Intraorbital tumor. The many types of intraorbital tumors also can present with proptosis, paralytic or restrictive ocular movements, and vision loss, especially if they involve the orbital apex. However, orbital tumors typically grow slowly over time; thus, their clinical features are usually more insidious. Corkscrew conjunctival vessels are not found with intraorbital tumors.

Vertebral-venous fistula. A case report describes a vertebral-venous fistula that presented with ocular symptoms attributable to a carotid cavernous fistula, but angiography demonstrated a vertebral-venous fistula. Although extremely rare, it should be kept in mind if carotid angiography is normal in a patient with symptoms consistent with a carotid cavernous fistula (34).

Indirect (“dural”) carotid-cavernous fistulas. In addition to the above differential diagnoses, indirect carotid-cavernous fistulas initially may be confused with simple ocular inflammation including blepharitis, conjunctivitis, episcleritis, or scleritis. However, indirect carotid-cavernous fistulas are not associated with discharge and will not resolve with topical antibiotics or anti-inflammatory medications. The conjunctival or scleral injection associated with these ocular inflammatory diseases can be discriminated from the typical radial, corkscrew vessels of carotid-cavernous fistulas. Lastly, cranial nerve palsies and other intracranial manifestations (eg, headache or transient ischemic attacks) occur frequently in indirect carotid-cavernous fistulas but not in these ocular inflammatory conditions.

Diagnostic workup

The key to diagnosing a carotid-cavernous fistula is to suspect its presence based on the history and physical examination. Direct carotid-cavernous fistulas present with dramatic orbital signs, such as severe head trauma and often with basal skull fractures. It is important to ask the patient about a bruit. Patients often describe the bruit as a whooshing noise or a heart beat in their head. A stethoscope, placed over the eye or at the temple, can be used to seek out this rather specific sign. Arterialization of the conjunctival vessels occurs more frequently in carotid-cavernous fistulas compared to other orbital diseases. Indirect carotid-cavernous fistulas, especially those of the posteriorly draining variety, are more difficult to diagnose by history and physical examination alone. One should still suspect them in any middle-aged to elderly woman presenting with a spontaneous red eye, chemosis, isolated or multiple cranial nerve palsy, and mild orbital congestion (60). Misdiagnosis is common. A retrospective series of 24 patients with carotid cavernous fistula identified 2 patients misdiagnosed with cluster headache, 1 as an orbital fracture and 1 as thyroid eye disease (07).

Another office procedure that may help diagnose a carotid-cavernous fistula is measuring the ocular pulse amplitude. The ocular pulse amplitude is the difference between the maximum and minimum intraocular pressure during the cardiac cycle and can be determined by tonometry. One study showed an ocular pulse amplitude of more than 1.6 mmHg to be 100% sensitive and 93% specific in distinguishing carotid-cavernous fistulas from normal subjects or those with other orbital diseases. However, direct and indirect carotid-cavernous fistulas cannot be distinguished from each other by the ocular pulse amplitude (32). The ocular pulse amplitude has been shown to normalize after closure of the fistula by endovascular embolization (32; 47).

The radiographic evaluation of a carotid-cavernous fistula will likely begin with noninvasive imaging such as CT, CT angiography, MRI, MR angiography, or transcranial Doppler ultrasound. A CT scan is usually the initial imaging study that is performed in patients with acute head trauma and may heighten the clinician’s suspicion for the presence of a carotid-cavernous fistula. In the case of direct carotid-cavernous fistulas, a CT scan will show proptosis of the involved globe, expansion of the cavernous sinus and superior ophthalmic vein, possible enlargement of the extraocular muscles, and any suspicious skull fractures (79). Magnetic resonance imaging has similar diagnostic utility to CT in finding signs suggestive of a carotid-cavernous fistula, but MRI is less sensitive in localizing bony fractures. An abnormal cavernous sinus flow void is a sign specific to MRI in the diagnosis of carotid-cavernous fistulas. In 1 study of indirect carotid-cavernous fistulas, CT found a dilated superior ophthalmic vein in all involved sides whereas MRI found this sign in 9 of 12 fistulas. MRI also found an abnormal flow void in the cavernous sinus in 11 of 12 patients with indirect carotid-cavernous fistulas (95).

For direct visualization of a carotid-cavernous fistula, CT angiography is increasingly being used in the initial localization and characterization of carotid-cavernous fistulas (23). In 1 study, CT angiography was found to be equivalent to traditional angiography and superior to MRA in determining the size and location of carotid-cavernous fistulas. This advantage was even greater in diagnosing direct carotid-cavernous fistulas involving the C4 segment of the internal carotid artery (16). Benson and colleagues reviewed 18 consecutive patients with a carotid-cavernous fistula and CTA imaging to find the most sensitive and specific imaging features of the carotid-cavernous fistula (09). Early enhancement and dilatation of the superior ophthalmic vein had 100% sensitivity. Arterial-phase contrast in the cavernous sinus had 88.9% sensitivity. 4D CTA is being increasingly used to diagnose neurovascular problems. MRA also can directly visualize a carotid-cavernous fistula but is less likely to delineate the small arterial feeder vessels or cortical venous drainage routes compared to traditional angiography (83). A report suggested that 4D flow MRI may be useful in providing a quantitative analysis of flow velocity and volume using time-resolved phase-contrast MRI for preoperative evaluation of a direct carotid cavernous fistula (66).

Other noninvasive imaging modalities are transcranial, transorbital color Doppler, and carotid duplex ultrasonography (18; 19). These techniques can measure the flow characteristics of both indirect and direct fistulas preoperatively. They may be even more important in helping to diagnose indirect carotid-cavernous fistulas when CT and MRI miss the signs of these fistulas. For example, results from a color Doppler ultrasound might help the clinician decide whether or not to perform invasive conventional angiography. Additionally, Doppler ultrasonography can give information regarding the collateral capacity of the circle of Willis before an intervention is undertaken as well as to judge whether an intervention was successful (48). Orbital color Doppler ultrasound (OCDUS) is another noninvasive test that, if negative, can rule out an anterior-draining carotid-cavernous fistula (96).

The gold standard for the diagnosis of both direct and indirect carotid-cavernous fistulas is catheter angiography, which determines the exact location of the fistula and delineates the feeder vessels and drainage pattern before treatment. If the cavernous sinus is draining superiorly into the cortical veins, the fistula should be treated emergently to prevent intracerebral hemorrhage and possible hemiparesis. Selective angiography also details the route (ie, transarterial or transvenous) by which the fistula can be closed, whether the fistula is being fed bilaterally, and the capacity of the circle of Willis (25). A common regimen that answers these questions includes selective injections of both internal carotid arteries and external carotid arteries, the ipsilateral vertebral artery, the ipsilateral ascending pharyngeal artery, middle and accessory meningeal arteries, and proximal and distal internal maxillary arteries (58). In a retrospective study of only 8 patients, using the DSA-Dynavision in treatment planning allowed more precision in their treatment and reduced the time of treatment (10).

Management

The definitive treatment of any arterial-venous fistula is obliteration of the fistulous connection. There are 4 methods of treatment of carotid-cavernous sinus fistulas: conservative, endovascular, stereotactic radiation, and surgical. The type of the fistula often determines the treatment of choice. It is important to remember that there still are no randomized clinical trials evaluating the treatment or outcomes of carotid-cavernous sinus fistulas.

Closure of indirect fistulas may occur spontaneously or after diagnostic angiography alone in as many 60% (76; 79). Indirect fistulas may respond to conservative treatment with a trial of carotid compression, which promotes thrombosis (40; 36; 46). The patient is asked to manually occlude the affected carotid with the opposite hand for 10 seconds 3 times an hour while awake. The occlusion time is gradually increased: After the second day, the patient compresses the artery for 15 seconds 3 times an hour for 2 days, then 20 seconds 3 times an hour for 2 days, then 25 seconds 3 times an hour for 2 days, then 30 seconds 3 times an hour. This method is less effective with direct fistulas, due to the high flow. Complete resolution of symptoms was reported in 21% to 34% of indirect carotid-cavernous fistula patients (40; 36; 46).

Conservative management alone may be successful even without thrombosis or closure of the indirect fistula. For example, 1 series found that 82% of patients (39 of 48) recovered or significantly improved with conservative treatment (27). However, a retrograde venous drainage pattern diagnosed by angiography raises the risk of intracerebral hemorrhage and mortality by 30% if the closure is delayed (36).

Endovascular treatment is the primary means of treating carotid-cavernous sinus fistulas with surgical intervention reserved for failures. Direct and indirect fistulas can be closed either by transarterial or transvenous routes, or a combination. The endovascular modalities include coils, stents, flow diverters, liquid embolics, particles, carotid sacrifice, detachable balloons, a vascular micro-plug, or combinations of these (08).

Direct carotid-cavernous sinus fistulas are usually closed by a transarterial coiling of the cavernous sinus. Sometimes the coils are prevented from falling back into the parent vessel with the help of a balloon or stent (64). In a series of 172 patients with direct fistulas treated with an endovascular procedure, the reported occlusion rate was 94%, with a 70% rate of preservation of the carotid artery (20). In the series, they had a 0.6% mortality and a 2.4% morbidity. Flow diverters may be used to treat direct fistulas (97; 90; 78). A systematic review showed that flow diverters used alone or as adjunct treatment may be useful, but long-term data are not available yet (86). Covered stents and flow diverters allow reconstruction of the vessel without occluding the sinus (49; 54; 70).

A direct fistula can be closed by is sacrificing the carotid artery. To predict safety of surgery, cerebral perfusion during carotid artery occlusion is tested. With the patient awake, under continuous clinical observation, the carotid artery is occluded for 15 minutes by balloon inflation. A nuclear perfusion study is performed at the end of the test. If the patient’s clinical and perfusion testing confirm robust collaterals, the artery may be sacrificed safely.

Indirect fistulas are more difficult to treat because there are usually multiple feeding vessels, which are often small, tortuous, and difficult to reach. The transvenous approach via the inferior petrosal sinus is the preferred route. Other options include the superior petrosal sinus, basilar plexus, pterygoid plexus, the facial vein, or even a direct puncture of the dilated superior ophthalmic vein or percutaneous access to the facial vein with ultrasound guidance (62; 58; 59; 100; 67; 24; 99; 56; 03). Coils, liquid embolics, or particles can be used to occlude the fistulous connection.

In a retrospective analysis of 267 patients, obliteration was accomplished in 86.5% (04). An exclusively transvenous approach had the highest obliteration rate of 86.9%. The transvenous approach using coils only had the highest likelihood of achieving a cure. Coiling alone had a lower complication rate when compared to liquid embolics with or without coiling.

Occasionally, a tortuous carotid artery limits the endovascular access to the fistula. An alternative modality to treat indirect fistulas in patients with contraindications for endovascular approach is endoscopic endonasal orbital apex decompression (84).

Surgical closure of the fistula involves posterior orbitotomy or craniotomy to gain access to the cavernous sinus (75; 79; 65; 85). In a series of 19 patients who failed endovascular treatment, 100% of the carotid-cavernous fistulas were closed with direct surgery. However, 3 patients (16%) required a graft to bypass the trapped intracavernous carotid artery, and approximately 26% of procedures required of sacrifice of the internal carotid artery or a bypass graft (94).

Targeted radiotherapy with gamma knife or X-Knife® triggers thrombosis and occlusion of the fistula. The closure rate for 22 indirect fistulas was 90.9%, with no recurrences during follow-up up to 14 years. However, improvement may take on average 2.4 months, and complete occlusion may take 7.5 months (05). This latency is consistent with other similar studies (77; 71; 21).

Complication rates depend on the radiotherapeutic modality. In 1 study, X-Knife® delivered higher radiation doses to the lens, optic nerve, optic chiasm, and brainstem compared to gamma knife (72). Stereotactic radiotherapy closed only a third of posttraumatic direct fistulas (05). In a retrospective study of 18 cases, at 2 years, 83% of patients had total obliteration of the direct carotid-cavernous fistula with stereotactic radiosurgery (73).

In patients with ocular compartment syndrome, emergent canthotomy/cantholysis may be performed by the ophthalmologist to decrease the intraocular pressure, as an emergency vision saving procedure (57).

Outcomes

Angiographic occlusion of the fistula predicts good outcome (45). Of 34 patients with both direct and indirect fistula, all showed complete recovery, and only 1 patient (3%) had a complication. Possible complications of treatment include intracranial hemorrhage or groin hematoma after endovascular treatment, new neurologic deficit, infection, recurrence, or endocrine dysfunction (38). The syndrome of inappropriate antidiuretic hormone secretion after Onyx embolization of bilateral direct carotid-cavernous fistulas was described in one patient (17).

Special considerations

Patients with impaired renal function. Renal dysfunction may aggravate following use of contrast during endovascular treatment. The risks and benefits must be weighed carefully and discussed with the patient and family.

Pregnancy

Women may develop indirect carotid-cavernous fistulas during pregnancy or post-partum due to increased risk of cavernous sinus thrombosis (93). In addition, the hemodynamic changes during pregnancy may explain reopening of a previously resolved direct carotid-cavernous fistula. The backflow to the cortical veins led to intracerebral hemorrhage and hemiplegia (55).