9.1 Goals

1. 
   

   Review clinical presentation of carotid-cavernous fistulas (CCFs) and discuss factors impacting diagnosis.

   
   
2. 
   

   Review indications for treatment.

   
   
3. 
   

   Review classification schemes and their role in determining treatment modality.

   
   
4. 
   

   Critically analyze current and future treatment options.

9.2 Case Example

9.2.2 Treatment Plan

Due to the concerns for a CCF, arrangements were made for a digital subtraction angiogram (DSA). A six-vessel DSA was performed, and it demonstrated fistulous filling of the left cavernous sinus during (Fig. 9.2a) selective left external carotid artery (ECA) injection, (Fig. 9.2b) selective right internal carotid artery (ICA) injection, and (Fig. 9.2c) selective left ICA injection. Venous outflow was via the superior ophthalmic vein to the angular and facial veins without evidence of retrograde cortical venous drainage.

Based upon the imaging findings, the recommendation was made for transvenous embolization of the fistula. The CCF was treated with transvenous coil embolization via the inferior petrosal sinus; coiling was directed from the confluence of the left superior ophthalmic vein and cavernous sinus in a retrograde manner to the posterior cavernous sinus. Complete obliteration of the CCF was achieved without complications.

9.3 Case Summary

1. 
   

   How can this carotid-cavernous fistula be classified?

   
   

   After the seminal 1985 publication, ¹ the Barrow classification became the most widely used scheme to describe CCFs. This classification system was based on angioarchitecture, and it categorized CCFs into four distinct types.

   
   

   Type A direct CCFs have a direct connection between the cavernous ICA and the cavernous sinus usually due to a rupture in the carotid wall. These CCFs are high-flow lesions and have a low chance of resolving without intervention. The high-flow characteristics and their rapidly progressive nature make type A CCFs more likely to lead to vision loss.

   
   

   Types B, C, and D are known as indirect or dural fistulas and arise from connections between the cavernous sinus and meningeal branches from the ICA (type B), meningeal branches from the ECA (type C), or both (type D). Type B, C, and D CCFs are less common than type A CCFs. For our patient, the imaging showed supply from bilateral internal and ipsilateral external carotid arteries (type D).

   
   
2. 
   

   Which patient factors aid in determining conservative management versus treatment?

   
   

   In 60 to 90% of cases, symptoms involve the orbit (conjunctival injection, chemosis, proptosis, glaucoma, diplopia, orbital hemorrhage, retro-orbital pain, and visual changes including vision loss). Less frequently, there can be progressive pain in the trigeminal distribution most commonly involving the V2 distribution. More rare presentations include intracranial hemorrhage, subarachnoid hemorrhage, epistaxis, or otorrhagia. CCFs can also present with headaches or tinnitus. While the decision to treat type A CCFs is straightforward, type B-D CCFs require more consideration given the greater likelihood of spontaneous resolution. While any threat to vision warrants treatment, patients with only headaches and/or tinnitus may require more nuanced discussions. In this instance, the patient had 6 months of intermittent diplopia, conjunctival injection, headaches, V2 sensory changes, and persistent tinnitus. She did not have hemorrhage, proptosis, elevated intraocular pressure, or vision changes; however, the symptoms were significantly impacting her quality of life. The treatment decision was elective and based on consideration of the risks, benefits, and alternatives.

   
   
3. 
   

   How did both the type of fistula play a role in determining treatment modality?

   
   

   There are multiple management options for CCFs: serial imaging, conservative treatment with manual external carotid compression, embolization, radiosurgery, and open surgery. However, endovascular treatment is the preferred treatment modality and includes the largest range of treatment options. Current endovascular modalities include detachable coils, liquid embolic agents (N-butyl cyanoacrylate [NBCA] and ethylene vinyl alcohol copolymer [EVOH]), covered stents, and flow diverting stents. Depending on the specific characteristics of the CCF, the lesion can be accessed via transvenous, transarterial, or combined approaches. Because this indirect CCF was fed from both ICA and ECA branches, we elected to proceed with transvenous coil embolization of the cavernous sinus.

   
   
4. 
   

   What manner of follow-up is appropriate for these lesions? Cross-sectional imaging of all types is limited in the ability to detect fistulous connections. Therefore, a follow-up DSA should be performed to confirm successful treatment. There is no definitive recommendation for the timing of a follow-up study, but treatment modality and symptomatology can influence the time frame. However, the most common strategy described in the literature is a 6-month follow-up angiogram with the option for an additional 1-year follow-up angiogram.

9.4 Level of Evidence

The current level of evidence for endovascular treatment of CCF is Level C due to the lack of randomized controlled trials or other data to support a higher level. This is not unexpected, given the relative infrequency of presentation and the changing treatment modalities over the past several decades. There are a large number of studies indicating benefit of endovascular intervention. However, these studies are almost entirely composed of cohort studies, retrospective analyses, and case reports. So while the level of evidence is low, treatment recommendations can be considered strong (Class Ha) due to the amount of published data and the anticipated natural history of CCFs.

9.5 Landmark Papers

Surgeries for pulsating exophthalmos have been performed since the early 1800s. If conservative measures failed, surgeons often proceeded to ligation of the common or internal carotid artery. Treatment modalities changed with the introduction of detachable balloon catheters by Serbinenko in 1974. ² The papers below give an abbreviated history on the classification of CCFs and how treatments have evolved since the 1980s. Beyond these select articles, there are many additional papers that have significantly contributed to the body of knowledge. ^{3 , 4 , 5}

Barrow DL, Spector RH, Braun IF, et al. Classification and treatment of spontaneous carotid-cavernous sinus fistulas. J Neuro-surg. 1985;62:248-256}

Previously, CCFs were classified according to three criteria: (1) traumatic or spontaneous, (2) high-flow or low-flow, and (3) direct or dural flow on angiography. These were applied in a variable manner in prior studies. Barrow et al posited that angiography actually demonstrated only four distinct types of CCFs, direct shunting and three variants of dural shunting. The classification proposed by Barrow et al effectively combined prior classifications and offered a four-part scheme. Type A CCFs with direct flow from the ICA generally have higher flow and are frequently traumatic in nature. Type B CCFs were determined to have indirect flow into the cavernous sinus from branches off the ICA, were slower flow states, and were most often spontaneous in origin. Type C CCFs had indirect flow from ECA branches and were spontaneous in origin. Finally, type D CCFs had indirect flow from both ICA and ECA branches and were more frequently spontaneous in origin. Treatment decisions could now be made based upon the CCFs classification.

This classification has become the most widely used schema for evaluating and determining treatment course. The effectiveness of the classification stems primarily from its assessment of the CCF’s angioarchitecture which has the greatest influence on the choice of therapeutic modality. The study also highlights the importance of cerebral angiography in the work-up of these patients. Despite the small sample sizes used to create the classification system, the results have held true in multiple subsequent studies.

Cognard C, Gobin YP, Pierot L, et al. Cerebral dural arteriovenous fistulas: clinical and angiographic correlation with a revised classification of venous drainage. Radiology. 1995;194(3):671-SSO. ⁶

In 1978, Rene Djindjian proposed a classification of purely meningeal fistulae into four types based on angioarchitecture and venous drainage. ⁷ Using this classification system, Cognard et al evaluated a consecutive series of 205 patients with the intention to validate and expand upon the original description. While their results closely mirrored those of Djindjian, they expanded and clearly defined the categories of indirect CCFs.

A Cognard type I fistula drained into the sinus with normal antegrade flow. Type II fistulas were subdivided into three categories: type Ha drained into a sinus with retrograde flow within the sinus; type lib drained into the sinus via retrograde flow through cortical veins with anterograde sinus drainage; and type Ila + b had retrograde drainage in both the cortical veins and the sinus. Type III fistulas drained directly into cortical veins with no demonstrated ectasia, while type IV fistulas demonstrated ectatic veins. Type V fistulas drained directly into spinal perimedullary veins. Thirty-three patients in their series had CCFs, and all were classified as type I, type lib, or typella + b.

Recently, Griauzde et al ⁸ argued against using the Barrow classification for indirect CCFs. They stated that direct CCFs are distinct entities from indirect CCFs and should be considered separately. The Cognard classification system was better able to identify high-risk fistulas requiring treatment by the identification of cortical venous reflux. In addition, Griauzde et al found that presentation with ocular symptoms had an association with Cognard groups Ila and Ila + b. Despite their arguments, however, the Barrow classification still remains the most widely implemented.

Debrun GM, Vinuela F, Fox A}, et al. Indications for treatment and classification of 132 carotid-cavernous fistulas. Neurosurgery. 1988;22(2):285-289. ⁹

Debrun et al applied the Barrow classification to their series of 132 CCFs and detailed the frequency of each type of fistula and demonstrated the utility of the Barrow classification to guide treatment.

The authors found that type A fistulas were the most common and most often associated with trauma. Of the 100 patients with type A CCFs, 95 were traumatic in origin. Ninety-two patients were treated with detachable balloons, and ICA patency was maintained in approximately 70% of the cases. Three patients required repeat endovascular treatment and four required open surgery after initial endovascular treatment. The complication rate was 30% and included transient and permanent oculomotor nerve palsies.

The authors also found that types C and D were dural arteriovenous fistulas with flow characteristics and angioarchitecture as described by Barrow et al—slow flow, spontaneous in origin, and originating from the branches of the ECA (type C) and a combination of the ECA and ICA (type D). Type D CCFs were more common than type C CCFs. No type B CCFs were seen in their series. With the exception of three type D patients who spontaneously resolved, all had endovascular treatment. All 4 type C and 25 of 28 type D patients were treated initially via particle embolization and/or liquid embolic agents through the ECA feeding vessels. This proved curative for the type C CCFs, but only 12 of 25 type D patients had complete obliteration even after several rounds of embolization. Several patients had transvenous balloon occlusion, and one required surgical exposure of the cavernous sinus for definitive fistula obliteration. These results reflected the more complex nature of type D CCFs.

Their series provided a wealth of corroborating information that was in accordance with the Barrow classification. The absence of type B CCFs in this group suggested that these lesions were uncommon. In addition, they demonstrated the ability of detachable balloons to treat type A fistulas without sacrifice of the ICA and the role of a multimodal approach in the treatment of persistent type D fistulas.

Meyers PM, Halbach W, Dowd CF, et al. Dural carotid cavernous fistula: Definitive endovascular management and long-term follow-up. Am J Ophthal. 2002;134(l):85-92. ¹⁰

In this retrospective analysis of 135 patients with indirect CCFs, 133 patients had endovascular treatment including 101 transvenous embolizations. They demonstrated the safety and efficacy of the transvenous approach with a 90% angiographic cure rate, 96% symptom improvement/resolution rate, and a 6% complication rate. With long-term clinical and angiographic follow-up, Meyers et al demonstrated the long-term efficacy of their treatments.

Ducruet AF, Albuquerque FC, Crowley RW, McDougall CG. The evolution of endovascular treatment of carotid cavernous fistulas: a single-center experience. World Neurosurg. 2013,80 (5):538-548. ¹¹

Ducruet et al reported managing 100 direct and indirect CCFs over an approximately 17-year period. In this timespan, endovascular treatments went through significant changes including the removal of the detachable balloon from the US market in 2004. These changes led to the use of coils as the mainstay for embolization. The data reflected both the changing treatments and the improvement in outcomes as endovascular techniques and tools evolved.

Detachable balloons were used to successfully treat 18 of the 42 type A CCFs. An additional 12 patients were treated with coil embolization. While ICA occlusion was necessary in 50% of the detachable balloon cases, only 25% of patients treated with coil embolization required ICA occlusion. The transvenous approach was used when either the transarterial approach was not possible or the transarterial approach failed to treat the fistula.

Coil embolization from the venous side was performed with a transarterial balloon or stent deployed within the ICA for protection. They reported a 10% complication rate (one death) for treatment of type A CCFs with an estimated 90% of patients demonstrating resolution or improvement in their symptoms.

Indirect CCFs were treated via the transvenous approach for 42 of the 48 patients in their series; 41 coil embolization cases and one liquid embolic case. Six patients underwent transarterial treatment and an additional five had combined approaches. Liquid embolics were utilized (seven with NBCA and four with EVOH) with the transarterial approach. The complication rate for indirect fistulas was -8% (no deaths) and > 80% experienced improvement or resolution of their symptoms.

This paper demonstrates an institution’s evolution in treatment of CCFs highlighting the innovations in endovascular equipment and advances in technique. The authors transitioned from transarterial balloon occlusion and ICA sacrifice to transvenous. By the time of publication, the authors were utilizing endovascular treatments as a first-line therapy with open surgery reserved for salvage therapy.