Discussion

Transvenous Onyx embolization has been reported in small cases series and case reports in a variety of pathologic scenarios. Most commonly, transvenous Onyx infusion has been applied in the treatment of indirect carotid cavernous fistulas.6–9 This technique typically involves the deposition of coils within the cavernous sinus followed by the slow infusion of Onyx to complete the occlusion of the venous drainage system.8 ,9 Obviously, this technique differs from the one we describe in which traversing the venous drainage pouch allowed catheterization of the ostium of an arterial feeder. The penetrating properties of Onyx enabled occlusion of multiple collateral arterial feeders without compromising venous drainage.

Transvenous occlusion of the cavernous sinus is generally well tolerated. The risk of pulmonary embolism is minimal given the relatively small caliber of the veins draining the cavernous sinus, specifically the petrosal sinuses and the ophthalmic veins. Complications have been reported, however, and include cranial neuropathies, bradycardia, asystole, and unintended migration of the agent into the carotid artery.3 ,6 ,8–11 Cranial neuropathies are thought to occur either from compression of the nerves from the embolic mass or chemical irritation from the dimethyl sulfoxide contained within the Onyx preparation.8–10 Similarly, compression or irritation of the trigeminal nerve can produce bradycardia and rarely even asystole through a vagal reflex arc.8–10 ,12 Despite these potential complications, Onyx embolization in this setting offers the advantages of shorter treatment times and greater penetration into the interstices of the cavernous sinus, thereby increasing the likelihood of cure.6

Using a technique similar to ours, Kessler et al5 reported the transvenous to arterial treatment of grade 3 or 4 AVMs in five patients. The authors advised employing a transjugular approach.5 Their rationale for this approach, which we support and employed in the management of our patient with an ethmoidal DAVF, was that this allowed the safe retention of the microcatheter within the jugular vein in cases with substantial reflux.5 The fragile nature of the veins draining an AVM would likely predispose to hemorrhage if an adhesed microcatheter was forcibly extracted. In addition, if a transfemoral approach were used, the retained catheter would traverse the right atrium and expose the patient to the risks of cardiac arrhythmias and thromboembolic complications. The concern over catheter retention and vessel rupture would be diminished with use of a detachable tip microcatheter. Unfortunately, while these catheters are available and widely used in Europe, they are not yet approved for use in the USA.

The effect of catheter adhesion is less of a concern in cases of DAVFs involving the major dural sinuses. In this setting, the dural leaves are substantially thicker than cerebral veins, thus making it far less likely to cause a rupture during catheter extraction. We employed the transjugular approach in the patient with the ethmoidal DAVF because we traversed a cortical vein, which is presumably more delicate than a dural sinus, in order to catheterize the ostium of a feeding artery.

Kessler et al5 argued that the transvenous to arterial approach was optimal in cases with favorable venous architecture or when the transarterial approach was not feasible. Using this approach, they cured four of five patients and substantially reduced the nidus in the last patient.5 Nonetheless, the potential morbidity of this procedure cannot be underestimated. If curative embolization is not accomplished and venous drainage of the AVM is compromised, the risk of subsequent hemorrhage is high. In DAVFs, however, occlusion of the venous drainage is less often detrimental and frequently curative.

Similarly, Dowd et al13 described a technique in which they transvenously catheterized a pericallosal arterial feeder in a 2-year-old patient whom they had previously treated in the perinatal period for a choroidal VOG malformation. The authors did not employ liquid embolysates, choosing instead to coil an enlarged feeding artery.13 While the technique is similar to ours, specifically, navigating through an enlarged VOG, the pathologic entity is not. In their case, the authors addressed a single fistulous feeder to a choroidal malformation. In our case, the use of Onyx allowed collateral embolization of numerous arterial feeders to a true VOG AVM.13

Lin et al14 reported an interesting case in which they cured a purely intraorbital arteriovenous fistula through a transvenous approach. In this case, the authors navigated a microcatheter into a branch of the superior ophthalmic vein and were able to push Onyx in a retrograde fashion to the site of the arteriovenous fistula.14 This technique allowed successful penetration of the fistulous site and resulted in a cure. This report differs from ours as the authors did not attempt, or need to attempt, catheterization of the ostia of feeding arteries.14 The flow properties of Onyx enabled them to cure the fistula from a distal catheter position. In our cases, this technique would have resulted in occlusion of the venous drainage system.14

Finally, Dalyai et al15 reported transvenous Onyx embolization of a cirsoid scalp AVM. In this case, the authors inflated a balloon in the external jugular vein as a means of preventing pulmonary embolism and improving retrograde flow of Onyx into the draining venous pouch.15 Again, this technique differs from ours in that the venous drainage system, rather than the feeding arteries, was occluded.

Our series highlights the efficacy of Onyx on a number of fronts. Importantly, this technique adds to the very limited endovascular treatment options for the management of ethmoidal DAVFs. These lesions are typically supplied by ethmoidal branches of the ophthalmic artery. Superselective catheterization of these small and tortuous branches is often impossible. Even when catheterization is feasible, distal access to a site close to the fistulous communication is exceedingly difficult to achieve. Add to this that embolization through these feeding vessels increases the risk of unintended embolization of, or reflux into, the central retinal artery. Our technique is ideal when there is a sizeable and non-tortuous draining cortical vein. Navigation through the sinuses is straightforward, and a transjugular approach decreases the morbidity of catheter retention and negates concerns over adequate catheter length.

Transarterial Onyx infusion has recently been described as an effective means of treating complex DAVFs.16 Nonetheless, microcatheterization of a position proximal to the fistulous site is not always possible nor does Onyx always penetrate distally enough to be curative.16 The transvenous to arterial technique offers an effective alternative in select cases. Both DAVFs treated and cured in our series were characterized by a recipient venous pouch parallel to the affected transverse and sigmoid sinuses. Navigating from the sinus into the pouch provided a stable position for catheterization of an arterial ostium. Such cases, in which a parallel pouch or a large draining vein allow transvenous navigation into an arterial ostium, are optimal for the transvenous to arterial infusion technique. Presumably, in DAVFs that drain directly into the transverse and sigmoid sinuses and do not have a parallel pouch, stable transvenous to arterial catheterization would be more difficult to achieve and would raise the probability of refluxing Onyx into the normal sinus.

In the case of the VOG AVM, however, we were forced to catheterize an arterial ostium directly opening into an enlarged VOG. An attempt to navigate a microcatheter primarily to this site was not feasible given its lack of stability and navigability within the high flow VOG. By employing a DAC within the straight sinus, a stable platform was constructed that allowed successful microcatherization through the VOG pouch. This technique may also be useful in DAVFs that drain directly into a normal sinus. Successful embolization into multiple arterial feeders in this case likely reduced the hydrodynamic pressure within the VOG and produced an improvement in the patient's hydrocephalus.

Despite the advantages of the transvenous to arterial technique, a number of potential complications must be considered. Reflux into normal cortical veins or dural sinuses can compromise normal cerebral venous drainage and produce stroke or hemorrhage. Theoretically, aggressive transvenous embolization can produce symptomatic pulmonary embolization. Finally, propagation of Onyx into arterial feeders to the cranial nerves or cerebrum can occur with potentially devastating neurological sequelae.

Once again, the transvenous to arterial technique offered an effective treatment option in our patient with a symptomatic VOG AVM. In her case, prior transarterial embolization had been performed but was no longer feasible given the diminutive size and tortuosity of the remaining arterial feeders. The transvenous approach resulted in a further decrease in the size of her nidus and stabilized her symptoms. Presumably, this retrograde means of infusing Onyx could also be applied in patients with choroidal VOG malformations.