complications and recommend the simultaneous compression of the eye bulb to avoid glue migration into the SOV. Shaibani et al. (2007) demonstrated that NBCA injection into a small compartment of the CS may close the AV shunt when an intracavernous stenosis does not provide catheter positioning for proper coil deployment.

Wakhloo et al. (2005) reported the largest experience so far with 14 patients; six treated with NBCA alone, seven with a combination of NBCA and coils, and one with a combination of transarterial PVA injection and intravenous NBCA. The authors achieved complete cure in all cases, observing two technical complications (14%) and one transient 6th nerve palsy (7%). There was one inadvertent glue migration into the MCA, another into the SOV; both without clinical sequelae. Intravenous glue injection should be performed with the greatest of care to avoid uncontrolled migration into the normal venous circulation or retrogradely via ILT and MHT into the carotid lumen, causing stroke. The latter has been observed by Meyers et al. (2002) who injected glue under pressure exceeding pressure in the small dural branches.

More recently, a modified type of acrylic glue (GlubranTM, GEM, Viareggio Italy) has become available that consists of a monomer NBCA and a monomer MS (owned by GEM). It has a lower thermal polymerization temperature than NBCA and is thought to be better controllable (see case Illustration VII). Preliminary experience in the treatment of brain AVM and DAVFs has been reported in one single study so far (Raffi et al. 2007).

8.4.2.4

Ethylene-Vinyl Alcohol Copolymer (OnyxTM)

The use of the liquid, non-adhesive embolic agent OnyxTM, (EV3, Irvine CA), mixed with dimethyl sulphoxide (DMSO) and tantalum (Jahan et al. 2001), was initially targeted for brain AVMs (Weber et al. 2007a,b; Song et al. 2007; van Rooij et al. 2007) and has only recently been extended to DAVFs (Suzuki et al. 2006; Toulgoat et al. 2006; Cognard et al. 2008; Rezende et al. 2006; Nogueira et al. 2008).

Regardless of its location, DAVFs have been considered the most difficult lesions to treat, especially by transarterial approach, because of their often elongated, tortuous ECA feeders, preventing a distal catheterization for effective glue injections. Transarterial embolization with NBCA has been

one main treatment modality for treating cranial DAVFs, but its efficacy remains limited to cases with favorable anatomy, in which either distal or wedged positions were obtainable, allowing for deposition of glue downstream at the desired location. The dependence on highly experienced operators for effective and safe use of liquid embolic agents has been widely overcome with the introduction of Onyx (Cognard et al. 2008).

After proper training, the flux of Onyx is, in general, easier to control, shortening the learning curve for its optimized handling. The anatomical results currently obtainable with Onyx appear superior to what can be accomplished with NBCA. Creating a so-called “embolic plug” (“reflux-hold-reinjection” technique, “reflux and push technique”) mitigates too much proximal reflux, while the embolic material penetrates slowly antegrade to the targeted vascular area. After casting the nidus itself, retrograde (“transnidal”) filling of adjacent feeding pedicles is possible, allowing for complete obliteration, not only of the nidus but also of a complete arterial network surrounding the AV shunt. Thus, the recruitment of new feeding vessels that form after partial or incomplete obliteration is minimized and will likely improve long-term results of embolizations in DAVFs, including DCSFs. Similarly, a direct endovenous application of OnyxTM is easier to control than that of acrylic glue, and can be used as an adjuvant technique to coils (Suzuki et al. 2006; Toulgoat et al. 2006; Rezende et al. 2006; Nogueira et al. 2008; Warakaulle et al. 2003; He et al. 2008; Lv 2009).

Toulgoat et al. (2006) treated six patients with DAVFs in various locations with OnyxTM 18, accomplishing complete occlusions in all patients with catheterization of a single feeder and one injection. Nogueira et al. (2008) reported on 12 consecutive patients with DAVFs embolized with OnyxTM 18 or a combination of OnyxTM 18 and OnyxTM 34, achieving complete obliteration in 11 patients (91.7%) with a total of 17 procedures. The authors observed no significant morbidity or mortality and saw one recurrence that required retreatment. Cognard et al. (2008) achieved complete closure in 80% of their patients (n=30), observing clinical complications in two (6.6%).

The combined use of coils and Onyx for endovenous occlusion of DCSFs in three patients has been described by Suzuki et al. (2006). The authors achieved complete occlusion and considered the controlled and excellent penetration of Onyx to be

superior for blocking the intricate communications in these fistulas.

Arat et al. (2004) performed transvenous injection of Onyx for treatment of a dural carotidcavernous fistula following an unsuccessful embolization using detachable coils and liquid adhesive agents resulting in complete resolution of symptoms after 3 month. The intracavernous injection was performed using a total of 0.6 ml of 8% ethyl-vinyl- alcohol-copolymer (Onyx 34) in a single injection casting not only of the CS, but also part of the SOV (see Fig. 3 in Arat et al. 2004). The latter should probably be avoided, since it may cause SOV thrombosis with possible loss of vision, the preservation of which is the goal of EVT in DCSFs.

A number of disadvantages and limitations should be considered when using this material (Suzuki et al. 2006; Jahan et al. 2001) including:

It requires DMSO-compatible microcatheters

Angiotoxicity with angionecrosis or vasospasm

Microcatheter retention due to entrapment

Costs exceed that of NBCA (in certain geographic regions, except US)

Discomfort in patients when treated without general anesthesia

Another, more general issue related to the use of OnyxTM is that the relative ease of control may lead to an underestimation of its potential of untoward penetration into venous compartments or exits that should not be occluded. In particular, within the CS, the degree of occlusion of only a selected compartment or all connections to afferent and efferent veins, and overfilling of the SOV are difficult to predict.

Case Illustration IX depicts how elegantly Onyx™ can be used in a DCSF with only minimal arterial supply through the ipsilateral AMA and MHT, when supported by indirect flow control. A balloon placed across the MHT and inflated during the Onyx injection into the AMA pedicles allows targeted deposition of Onyx™ with the CS. This prevents distal migration of the embolic agent into the SOV or reflux into the ICA.

Transarterial embolization of DAVFs may be accompanied by clinical complication due to proximal reflux causing CN deficits, or overinjection and subsequent extensive venous thrombosis (Cognard et al. 2008). Recent literature provides an increasing number of smaller series reporting the successful use of Onyx™ for both transvenous and transarterial embolization of DCSFs (Ong et al. 2009; Lv et al. 2008; Gandhi et al. 2009; Elhammady 2009;

Bhatia 2009; He 2008). Although the achieved high rates of technical success and occlusion are promising, the use of OnyxTM in CSFs is not without potential hazards.

He et al. (2008) treated six patients using Onyx and coils observing two transient 6th nerve palsies after TVO (33%) and one transient facial nerve palsy (17%) after TAE.

In another series, complete occlusion was achieved in 11 patients undergoing TVO using a combination of coils and Onyx (Lv et al. 2009). Two patients (18%) developed a bradycardia during the DMSO injection; transient or permanent CN palsies were not observed.

Bathia et al. (2009) reported five patients undergoing TVO using Onyx-34 and achieved complete occlusion with resolution of clinical symptoms in all. One patient (20%) showed persistent 6th cranial nerve palsy that resolved after 3 months. The risk of potential reflux into cavernous ICA branches is emphasized as a major disadvantage. It can be minimized to some degree by using a higher viscosity (Onyx-34), and pausing the injection whenever reflux occurs.

Similarly, Elhammady et al. (2009) treated a group of 12 CSFs, with 11 DCSFs, using Onyx for arterial and endovenous injections. The authors used a combination of coils and Onyx in five and Onyx only in three patients for TVO, achieving complete occlusion in all. In three other patients, TAE was performed resulting in complete obliteration as well. The authors observed two transient CN palsies (18%) and one permanent facial nerve palsy (9%) in the patients undergoing transarterial embolizations. This relatively high morbidity may be caused by ischemia in the arterial territory supplying the cranial nerves in the CS, by aggravated CS thrombosis and swelling, or is possibly related to direct angiotoxic effects of DMSO.

These newer observations may tone down a bit the prevailing Onyx enthusiasm and illustrate several important aspects: First, caution is warranted before injecting DMSO and Onyx directly into the CS as it seems currently unclear to what degree the cranial nerve function may be affected. Second, the use of Onyx must be performed under the same rules used for injecting NBCA. This includes avoiding dangerous anastomoses and respecting the normal arterial supply to the cranial nerves. Third, larger series with long-term clinical and angiographic FU are needed to validate efficacy and procedural morbidity of Onyx embolizations (Wakhloo 2009).

Murugesan et al. (2008) recently reported a severe adverse pulmonary reaction following embolization of a cerebral AVM. The patient developed acute respiratory distress syndrome with hypoxemia following extubation, necessitating mechanical ventilatory support for 44 h. This unusual complication may be related to the excretion of DMSO through the lungs.

Increased radiation exposure due to prolonged injection time [mean injection time 45 min (Cognard et al. 2008)] is another controversial issue, but must be counterbalanced against long fluoroscopy time, often necessary for more distal catheter navigation and repeated sessions in the use of NBCA (Nogueira et al. 2008). In DCSFs, procedure time and radiation exposure may actually be reduced compared to time-consuming coil packing (Gandhi et al. 2009; Bhatia et al. 2009). Catheter entrapment may cause clinical complications (Carlson et al. 2007), but will likely be solved by technological advancements, such as the recent introduction of a detachable tip for the Sonic catheter from Balt (Tahon et al. 2008).

Overall, the experience with Onyx is still limited but encouraging, as it allows for better and more controlled penetration of the complex network of an AV shunt as compared to NBCA, and can be considered a valuable adjunct to the EVT armamentarium for the management of DCSFs, either alone or in combination with detachable coils.

It should be mentioned here that among liquid embolic materials, the use of alcohol has also been reported (Koebbe et al. 2003). In a group of six patients, intraarterial alcohol injection was performed under temporary balloon inflation in the ICA. Technical success was achieved in all cases, clinical improvement in five. One patient experienced worsening of her sixth nerve palsy due to CS thrombosis. Although known as highly effective for vascular occlusion, this technique carries a high risk of damaging the ICA endothelium as well as the cranial nerves traversing the CS, and thus should be used only as a last resort.

8.4.2.5

Stents and Covered Stents

The use of stents in the treatment of CSFs has been reported in a limited number of cases (Moron et al. 2005; Archondakis et al. 2007; Felber etal. 2004; Gomez et al. 2007), mainly in association with direct CSFs. Moron et al. (2005) demonstrated that preservation of the parent artery was achieved us-

ing stent-assisted coiling in five patients. The actual benefit in using stents lies in the possible reduction of the arterial inflow when deploying stents with a low porosity, or ideally, covered stents. Angiographic follow-up in eight patients with traumatic CCFs treated with covered stents (JoStent Coronary Stent Graft) demonstrated improved symptoms or complete regression in all. In two, a residual filling of the AV shunt was found at the end of the procedure. Six patients showed good patency of the carotid lumen, while one presented with an asymptomatic occlusion (Arachnondakis et al. 2007).

For effective use of covered stents, good wall apposition is crucial to fully seal the defect in the carotid wall of a CCF, or the TMH/ILT origins in a DCSF. If the stent is malapposed, the AV shunt may stay open (Lv et al. 2008). Care is necessary when further expanding the stent in a CCF, since an injured arterial wall provides little resistance. Thus, when a stent graft is over-inflated, additional damage could potentially enlarge the defect in the wall (personal observation in one case). Lv et al. (2008) observed the development of a complex AVF 9 months after placement of a covered stent for occlusion of a traumatic CCF.

The use of covered stents in dural cavernous sinus shunts has been described in one case by Kim et al. (2006). Residual slow flow directed toward pontomesencephalic veins occurred after transvenous coil packing, resulting in brain stem congestion associated with dysarthria. Bilateral placement of a stent graft resulted in complete occlusion of the shunt. Although neurological deterioration after TVO certainly justifies such an aggressive approach, a number of potential disadvantages have to be considered.

First, no covered stent dedicated for intracranial circulation is currently available. The stiffness of the JoStent coronary graft is an inherent limitation that may not allow easy navigation and can cause spasm or even dissection (Archondakis et al. 2007). Second, as with any stent placement, dual antiplatelet therapy (aspirin and clopidogrel) 2–3 days prior to the stent placement and 2–3 months post stenting, then aspirin only for 12 months or even indefinitely, is recommended. The required large size of the guiding catheter (7 or 8 F) in combination with agressive anticoagulation may cause additional complications at the puncture site.

Third, the shortand long-term patency of these stents is unknown. The PTFE layer may cause acute inflammation and ingrowth of fibrous connective tissue (Geremia et al. 1997; Link et al. 1996). On