Direct carotid-cavernous fistula: A complication of, and treatment with, flow diversion

Abstract

Direct carotid-cavernous fistulas (CCFs) are rare complications of flow diversion and have typically been documented in a subacute time frame after treatment. We present the first reported case of an intraprocedural direct CCF that developed immediately after flow diversion for treatment of a symptomatic paraclinoid right internal carotid artery aneurysm with a neck involving the cavernous segment. Endovascular treatment of such direct fistulas typically involves either transarterial obliteration of the fistulous site or transvenous embolization of the cavernous sinus. Our case was successfully treated with further immediate flow diversion without additional transvenous intervention. There are few reports on the use of flow diversion for treatment of such direct CCFs, and in all but one of these cases, flow diversion was combined with concomitant transvenous embolization. Thus, the presented case is not only the first reported case of an immediate CCF after flow diversion, but it is also only the second reported case of a direct fistula to be successfully treated using solely flow diversion, without additional transvenous intervention. We review the literature of direct CCFs after flow diversion, the pathophysiology of development of CCFs after flow diversion, the literature on treatment of CCFs with flow diversion as well as all other current treatment options.

Introduction

Flow diverters such as the Pipeline embolization device (PED; Covidien, Mansfield, MA, USA) are a new class of low-porosity, self-expanding, stent-like devices, specifically designed for the endovascular management of select intracranial aneurysms previously considered untreatable by conventional techniques.

As experience with flow diversion increases, interventionalists have gained a greater understanding of the complications of flow diversion such as vessel injury, thromboembolic events and perforator branch occlusions.¹ A rare but known complication following flow diversion is delayed aneurysm rupture.¹ If the target lesion involves the cavernous segment of the internal carotid artery (ICA), delayed aneurysm rupture or parent vessel injury can lead to an abnormal communication between the ICA and the cavernous sinus, resulting in a carotid-cavernous fistula (CCF).¹ Direct (Barrow type-A) CCFs typically result from traumatic laceration of the cavernous ICA, or traumatic avulsion of cavernous ICA branches. Spontaneous direct CCFs occur in the context of angiodysplasia, or ensue after rupture of a cavernous-carotid aneurysm.²

The development of a direct Barrow type-A CCF after flow diversion for treatment of cavernous-carotid aneurysms is extremely rare, with only seven cases reported in the literature, all of which developed subacutely from three days to 110 days after treatment.^1,3–6 We present the first case of an intraprocedural direct CCF that developed immediately after flow diversion for treatment of a paraclinoid ICA aneurysm with a neck involving the cavernous segment.

Endovascular treatment of such direct CCFs typically involves either transarterial obliteration of the fistulous site or transvenous embolization of the cavernous sinus. Upon diagnosis of the direct CCF in the presented case, the fistula was successfully treated with further immediate flow diversion without additional transvenous intervention, which has been reported only once previously.

Thus, the presented case is extremely unique not only because it is the first reported case of an immediate, intraprocedural CCF after flow diversion, but it is also only the second reported case of a direct CCF to be successfully treated using solely flow diversion, without additional transvenous intervention. We present this case not only as a review of a rare complication of flow diversion, but also as a review of flow diversion in the treatment options of such lesions.

Case presentation

A 69-year-old woman with a history of hypertension and two years of progressive right-sided vision loss due to an unruptured paraclinoid right ICA aneurysm presented with worsening left-sided vision over several months. On exam, she had no light perception in the right eye, and 20/60 visual acuity in the left eye. Pupillary testing revealed a right relative afferent pupillary defect with a normal and brisk left pupillary response to light. Extraocular movements were normal in both eyes. Magnetic resonance imaging (MRI) of the brain and digital subtraction angiogram (DSA) demonstrated a 12 mm × 12 mm paraclinoid right ICA aneurysm with a 7 mm neck that extended proximally to involve the cavernous segment of the ICA (Figure 1(a), (b)). The aneurysm had grown in size compared to DSA from two years earlier with worsened mass effect on the optic chiasm. Because of the worsening left-sided vision, the location and wide neck of the aneurysm, the decision was made to treat with flow diversion.

Figure 1.

(a) Magnetic resonance imaging (MRI) brain, coronal T2-weighted image shows large paraclinoid right internal carotid artery (ICA) aneurysm with mass effect on the optic chiasm. (b) Digital subtraction angiogram (DSA) shows wide-necked aneurysm. (c) Unsubtracted lateral view immediately after Pipeline embolization device (PED) #1 shows contrast stagnation in aneurysm, apposition of PED to parent ICA. (d) DSA of right ICA in lateral view immediately after PED #1 shows patent anterograde flow and no evidence of carotid-cavernous fistula (CCF). Subsequent DSA of right ICA in (e) frontal and (f) lateral views shows decreased anterograde flow and development of CCF with retrograde flow anteriorly through superior ophthalmic vein (arrows), inferiorly to pterygoid plexus (arrowheads) and clival plexus (open arrows).

The patient was treated with dual-antiplatelet therapy with aspirin 325 mg and clopidogrel 75 mg. On pre-procedural point-of-care platelet function testing using P2Y12 assay (VerifyNow; Accumetrics, San Diego, CA, USA), she was found to be nonresponsive to clopidogrel. Thus, ticagrelor 90 mg was started in place of clopidogrel for an additional five days. On the day of the procedure, repeat point-of-care platelet function testing showed adequate platelet inhibition.

During the procedure, a 90-cm Neuron MAX 0.088-inch inner diameter (ID) sheath (Penumbra, Alameda, CA, USA) was placed into the proximal cervical right ICA, through which a 115-cm Navien 5-French, 0.058-inch ID intermediate catheter (Covidien, Irvine, CA, USA) was placed in the petrous right ICA. A 150-cm Marksman microcatheter (Covidien, Irvine, CA, USA) was tracked into the distal supraclinoid ICA without difficulty and through this, a 4.5 mm × 18 mm first-generation PED was successfully deployed without excess manipulation or complication, with the distal end in the paraophthalmic segment, extending across the aneurysm neck, and with the proximal end in the cavernous ICA. Immediate post-PED DSA showed adequate apposition of the PED to the parent vessel, contrast stasis within the aneurysm and no evidence of vessel injury, extravasation or early venous filling (Figure 1 (c), (d)). After approximately 10 minutes, a repeat DSA showed early opacification of the right cavernous sinus, inferior petrosal sinus and superior ophthalmic vein, consistent with development of a direct CCF (Figure 1(e), (f)).

Given the fact that optimal access for flow diversion was already established, the decision was made to proceed with further flow diversion in attempts to exclude the fistulous site. Via the Marksman microcatheter, four additional first-generation PEDs (5 mm × 16 mm, 5 mm × 14 mm, 5 mm × 12 mm, 5 mm × 18 mm) were telescoped within the paraclinoid right ICA, across the aneurysm neck, and extended proximally within the cavernous segment of the ICA, with resulting decreased flow through the CCF (Figure 2(a), (b)).

Figure 2.

(a) Digital subtraction angiogram (DSA) of the right internal carotid artery (ICA) in (a) frontal and (b) lateral views after deployment of four additional Pipeline embolization devices (PEDs) shows improved anterograde flow and decreased filling of carotid cavernous fistula (CCF). (c) DSA on post-intervention day #1 of right ICA in lateral view shows continued improvement of anterograde flow and further decreased filling of CCF. Note decreased opacification of superior ophthalmic vein, pterygoid plexus and clival plexus. (d) Twelve-month follow-up DSA shows vessel reconstruction with resolution of CCF.

The patient was extubated and on post-interventional exam, her conjunctiva and sclera were white and quiet bilaterally without injection or chemosis. There was no palpable thrill or bruit on auscultation over either eye. Her visual acuity and extraocular movements were unchanged. Sensation in the V1–V3 distributions of the trigeminal nerves was normal bilaterally. Dilated fundus exam was normal bilaterally except for unchanged mild pallor of the right optic nerve. Her intraocular pressures (IOP) using Goldmann Tonometry measured 21 mm Hg in the right eye and 21 mm Hg in the left eye, and remained stable overnight.

The following day, the patient returned for anticipated transvenous embolization through the ipsilateral inferior petrosal sinus. However, repeat angiography showed markedly decreased filling of the CCF (Figure 2(c)). Given the patient’s lack of symptoms, stable IOPs and improving radiographic findings, the decision was made to forego transvenous embolization. Her IOPs continued to decrease and on discharge on post-intervention day #4, her IOPs measured 18 mm Hg in the right eye and 15 mm Hg in the left eye.

The patient was discharged on dual-antiplatelet therapy with aspirin and ticagrelor. At both two- and 12-month follow-up, she remained asymptomatic with stable ophthalmologic exam, stable left visual acuity and no clinical evidence of CCF. A 12-month follow-up angiogram showed progressive stasis of intra-aneurysmal flow, with complete resolution of CCF (Figure 2(d)).

Discussion

With the introduction of flow-diverting stent technology, neurointerventionalists have been able to treat an increasingly more difficult subset of cerebrovascular pathology that had previously been considered untreatable by conventional techniques. Along with this increased usage, a greater understanding of the complications associated with flow diversion has developed. Complications associated with the usage of flow diversion, and specifically PED, have been well documented.^1,7–9 A recent systematic review of 10 publications involving PED use in 414 patients with 448 cerebral aneurysms identified an overall procedural complication rate of 10.3% (46/447), with an intracranial vascular complication rate of 6.3% and procedural mortality of 2.2% (9/413).⁹ A meta-analysis of 29 studies performed by Brinjikji et al. including 1451 patients with 1654 aneurysms noted a procedure-related morbidity of 5% and mortality of 4%.¹⁰

Procedural complications of flow diversion include, but are not limited to, thromboembolic events and device occlusion, intracranial hemorrhage, perforator/side branch occlusion and vascular injury.¹ If the target lesion includes the cavernous ICA, vessel injury during stent deployment or delayed rupture of the aneurysm after flow diversion may lead to the spontaneous development of a direct CCF.

The development of a direct (Barrow type-A) CCF after flow diversion is extremely rare, and has been reported in only seven patients, all of which occurred subacutely, within 3–110 days after treatment (Table 1).^1,3–6 Of the seven reported cases of direct CCFs after flow diversion, four involved the PED and three cases involved the Silk flow-diverting stent (SFD; Balt Extrusion, Montmorency, France). Results from the Pipeline for Uncoilable or Failed Aneurysms (PUFS) Clinical Trial report one case of CCF 180 days after treatment of a cavernous segment aneurysm, which was determined to be “probably related” to PED deployment.¹¹ In their series of 137 intracranial aneurysms treated with PED, Park et al. report one direct CCF on short-term follow-up studies in a patient treated with a PED for a traumatic carotid cave/cavernous ICA pseudoaneurysm.¹ That patient was successfully treated with transvenous coil embolization of the fistula. Lin et al. and Mustafa et al. both report on direct CCFs that developed two to six weeks after flow-diverter treatment of symptomatic cavernous carotid aneurysms, which were all successfully treated using transvenous embolization.^3,4 Kulcsár et al. report two cases of direct CCFs following Silk flow diversion treatment of cavernous carotid aneurysms, one of which developed on postintervention day #3, and the other on day #110.⁵ Both patients were treated with parent vessel occlusion.

Table 1.

Reported cases of direct CCFs after flow diversion.

Study, year	Aneurysm type	Aneurysm treatment	Interval to CCF formation	Treatment of CCF
Lin et al., 20153	Cavernous left ICA aneurysm	PED	6 weeks	Transvenous coil embolization of CS
Lin et al., 20153	Cavernous left ICA aneurysm	PED	1 month	Transvenous coil embolization of CS
Park et al., 20151	Traumatic carotid cave/cavernous ICA pseudoaneurysm	PED	Short term	Transvenous coil embolization
Becske et al. (PUFS), 20136	Cavernous ICA aneurysm	PED	180 days	Unknown
Kulcsár et al., 20115	Cavernous left ICA aneurysm	Silk	3 days	Parent artery occlusion
Kulcsár et al., 20115	Cavernous right ICA aneurysm	Silk	110 days	Parent artery occlusion
Mustafa et al., 20104	Cavernous right ICA fusiform aneurysm	Silk	2 weeks	Transvenous coil embolization

CCF: carotid cavernous fistula; ICA: internal carotid artery; PED: Pipeline embolization device; PUFS: Pipeline for Uncoilable or Failed Aneurysms; CS: cavernous sinus.

Our case is the first reported direct Barrow type-A CCF to develop immediately after deployment of the PED. Both the positioning of the Marksman microcatheter and the deployment of the PED proceeded without excess manipulation or complication. An immediate post-PED deployment angiogram showed appropriate and expected stagnation of flow into the aneurysm and no evidence of vessel injury, dissection, extravasation or CCF (Figure 1(c), (d)). It was only a subsequent DSA that showed the development of the CCF. Thus, the CCF was not due to a technical or mechanical vessel perforation during device deployment or recapture, but rather to an inherent change in the underlying pathophysiology of the parent ICA and/or aneurysm wall itself. Although the dome of the aneurysm was predominantly intradural, the aneurysmal neck extended proximally to involve the distal cavernous ICA. Thus we hypothesize that a pathophysiological change at the interface between the cavernous ICA and the aneurysm neck ultimately caused the CCF, rather than rupture at the aneurysmal dome, which would have potentially led to a traditional subarachnoid hemorrhage. In the event of subarachnoid hemorrhage in this situation, a similar treatment course would have been ventured; the priority would have been to continue embolization across the neck of the aneurysm as quickly as possible using additional PEDs.

The incidence of delayed spontaneous aneurysm rupture after treatment with a flow diverter ranges from 0.6% to 1%.^3,12 The two hypothesized pathomechanisms of delayed aneurysmal rupture after flow diversion involve development of intra-aneurysmal thrombosis and acute changes in flow hemodynamics.⁵ After flow diversion, intraluminal thrombus formation within the aneurysm sac and mural thrombus along the aneurysm wall is theorized to cause intimal damage and intramural inflammation with further degeneration of the aneurysm wall.^5,12 Aggregation and infiltration of leukocytes along the intimal surface leads to reverse-remodeling of the vessel wall involving a combination of increased elastolysis, adventitial immunoinflammation and increased secretion of proteases.^5,12 This local cascade effect may lead to further chemical degradation and weakening of the aneurysm wall and subsequent rupture.

A change in parent vessel and intra-aneurysmal hemodynamics has also been hypothesized to contribute to delayed rupture after flow diversion.^5,12 Flow diversion may lead to a sudden change in intra-aneurysmal hemodynamics with increased stress to areas that were not previously exposed to strain, which may cause rupture. Additionally, the inherent flow diversion into the higher resistance parent artery pathway, cerebral autoregulation leading to higher-pressure gradients and changes in the parent artery configuration may also contribute to delayed rupture.¹² Although the underlying etiology of rupture in our presented case is impossible to determine, we hypothesize that a change in the hemodynamic pattern at the interface of the parent vessel and neck was the major contributor. Since the fistula originated at the level of the cavernous ICA, an abrupt change in the flow-diversion hemodynamics at the interface of the parent vessel and the proximal PED may have been the inciting factor, rather than inflammatory degeneration of the aneurysmal dome, which would have otherwise caused subarachnoid hemorrhage. Further, intraluminal thrombus causing a cell-mediated, reverse-remodeling of the aneurysm wall may have a greater temporal component associated with it, and may be the predominant contributing factor in late subacute ruptures, rather than in an immediate rupture, as in our case.

Traditional management of direct CCFs resulting from rupture of an untreated CCA has consisted of transarterial obliteration of the fistulous site. The goal of treatment is to occlude the site of communication between the ICA and the cavernous sinus while preserving the patency of the ICA.³ Transarterial embolization of the cavernous sinus through the fistulous site using detachable coils or liquid embolic agents such as n-butyl cyanoacrylate (NBCA; Trufill, Codman, Raynham, MA, USA) or Onyx (ev3 Neurovascular/Covidien, Irvine, CA, USA), has become the mainstay transarterial option.¹³ During transarterial embolization, a temporary balloon or a self-expanding stent may be placed in the parent ICA across the fistulous site to protect the parent vessel and prevent coil migration or prolapse.

Other treatment options for treating CCFs include transvenous embolization, parent artery sacrifice and operative methods such as surgical ligation.^3,13 Transvenous occlusion of a direct CCF involves embolization of the cavernous sinus via a posterior approach through the inferior petrosal sinus or an anterior approach through the superior ophthalmic vein. Once access in the cavernous sinus is obtained, transvenous embolization of the cavernous sinus can be accomplished using detachable coils, hydrogel-coated coils or liquid embolic materials.³

Once the diagnosis of CCF was established in our case, various treatment options were considered including transvenous embolization and deconstructive treatment options. A transarterial approach through the fistulous site was not an option because the low porosity of the PED prevents transarterial access into the aneurysm or into the rupture site or fistulous connection. An immediate balloon test occlusion of the ipsilateral ICA was considered in order to evaluate for a possible vessel sacrifice. However, given the fact that the patient was already intubated, emergent neurophysiological monitoring was not available and optimal access for flow diversion was already established, the decision was made to proceed with further flow diversion in attempts to exclude the fistulous site.

Recently, the use of flow diversion for the treatment of direct CCFs has been reported, although only in five cases (Table 2).^14–17 In all but one of these cases, transarterial flow diversion was combined with concomitant transvenous embolization of the ipsilateral cavernous sinus. Thus, our case is only the second reported case of a direct Barrow type-A CCF to be successfully treated using solely flow diversion, without additional transvenous intervention. Nossek et al. describe a case of a direct CCF due to a ruptured giant left cavernous carotid.¹⁷ A total of three telescoping PEDs were deployed across the aneurysm neck, as well as subsequent coil embolization of the cavernous sinus. Pradeep et al. describe two cases of post-traumatic CCF, both treated with a combination of transarterial flow diversion across the fistulous site with concomitant transvenous coil and Onyx (ev3 Neurovascular/Covidien, USA) embolization of the cavernous sinus.¹⁶ Iancu et al. report a case of post-surgical direct CCF that was treated with transarterial coil embolization of the right superior ophthalmic vein (SOV) and cavernous sinus with balloon protection, subsequently followed by deployment of a Silk stent into the supraclinoid and cavernous ICA.¹⁵ Nadarajah et al. describe a case of a post-traumatic lacerated cavernous right ICA with subsequent CCF that was successfully treated with a telescoping construct of four PEDs.¹⁴

Table 2.

Reported cases of direct CCFs treated with flow diversion.

Study, year	Etiology	Presentation	Finding	Treatment	Outcome
Nossek et al., 201517	Aneurysm rupture	Acute left-sided ophthalmoplegia	Ruptured left cavernous carotid aneurysm with fistulous outflow via the left SOV and into the pterygoid venous plexi bilaterally	PED × 3 in left ICA; transarterial coiling of the aneurysm and left cavernous sinus	Complete symptomatic resolution; obliteration of the fistula with occlusion of the aneurysm at one-year follow-up DSA
Pradeep et al., 201516	Trauma	Left CN6 palsy, ptosis with a left lateral field cut, pulsatile tinnitus, exophthalmos of the left eye	Direct CCF of the left ICA with anterior and posterior venous drainage	PED × 2 in left ICA; Transvenous coil and Onyx embolization of left cavernous sinus	Persistent left CN6 palsy, improved exophthalmos and left lateral field deficit at three months
Pradeep et al., 201516	Trauma	Left-sided exophthalmos, left CN3, 4, 6 palsy	High flow, direct left CCF	PED in petro-cavernous left ICA; Transvenous Onyx embolization of left cavernous sinus; PED × 2 in left ICA over next six months	Completely remodeled ICA and occlusion of the CCF at seven-month follow-up DSA
Iancu et al., 201515	Post-surgical	Right proptosis, chemosis, double vision, right CN3 palsy	Right ICA injury with CCF refluxing into right SOV	Transarterial coil embolization of right cavernous sinus and SOV; Silk stent in cavernous right ICA	Oculomotor symptoms immediately improved. Normal angiogram on one-year follow-up
Nadarajah et al., 201214	Trauma	Right orbital swelling and retro-orbital pain, right CN6 palsy	Transection of cavernous right ICA with CCF; partially thrombosed and grossly distended SOV	PED × 4 in cavernous right ICA	Complete resolution

CCF: carotid cavernous fistula; DSA: digital subtraction angiography; ICA: internal carotid artery; SOV: superior ophthalmic vein.

In our case, four additional PEDs were further deployed across the neck of the aneurysm as well as across the site of fistulous communication. Multiple PEDs were telescoped to provide high metallic coverage and low porosity in the regions of device overlap over the fistulous site. This multi-PED construct maximizes the benefits of flow diversion, and offers a premium scaffold for endothelial overgrowth.¹⁸ After the additional four PEDs were deployed, the CCF showed drastically decreased filling (Figure 2(a), (b)). The following day, the possibility of transvenous embolization was expected as the sump effect of the high flow fistula was anticipated to keep the communication open despite PED placement. However, there was markedly further decreased filling of the CCF (Figure 2(c)). Given the patient’s stable clinical status and IOPs, the decision was made to forego transvenous embolization to avoid further risk.

One of the major limitations of using flow diversion to treat patients with CCFs is the requirement for dual antiplatelet therapy, which may preclude the usage of these devices in acute settings.¹⁶ In our case, a PED had already been completely deployed, thus the need for dual-antiplatelet therapy was already obligatory. Despite the presence of a patent fistula, we felt compelled to discharge the patient on dual-antiplatelet therapy because of the increased risk for symptomatic thromboembolic events in the presence of multiple PED placements.¹⁹ Additionally, given the markedly decreased flow through the fistula on the following day and the patient’s stable clinical status, we were hopeful that the CCF would ultimately thrombose despite dual-antiplatelet therapy; a foregone conclusion that ultimately transpired at both two- and 12-month follow-up. In light of these challenges, some authors suggest a staged approach that initially uses traditional techniques for CCF occlusion, later combined with the synergistic effect of flow diversion.¹⁶

Direct CCFs are a rare complication of flow diversion and have typically been documented in a subacute time frame after treatment. Our case shows that even acute changes in parent vessel and intra-aneurysmal hemodynamics may cause an immediate CCF. Compared with the use of detachable balloons, coils and vessel sacrifice, the use of flow-diverting stents to treat such direct CCFs may prove to be safe and efficacious, especially in situations in which transarterial embolization through the fistulous site or transvenous embolization is not possible. Larger, controlled studies are necessary to determine the ultimate role of flow diversion in the treatment of such CCFs.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.