Summary

Complete understanding of the vascular anatomy of DCSFs has been compromised by the complexity of the supplying dural arterial network and by the nature of the cavernous sinus as a venous collector draining blood from cerebral and orbital venous circulation. Since its initial description, and despite numerous angiographic and cadaver studies, controversy concerning the true anatomic structure of the cavernous sinus (CS) persists to some extent still today. The CS is most likely not an unbroken trabeculated venous cavity, or a plexus of varioussized veins; rather it represents a complex venous compartment where numerous dural sinuses and veins converge to form larger venous spaces around the carotid artery, which could be termed caverns and whose angiographic appearance is influenced by hemodynamics and the existence of intracavernous thrombi.

More likely than the initially and erroneously assumed rupture of the thin-walled dural arteries, a spontaneous thrombosis within the cavernous sinus, seen in as many as 62% of DCSFs, is believed to play a major role in the etiology of the fistulas. Recanalization of thrombosed sinus compartments leads to functional enlargement of micro-AV shunts within the rich vascular network of the dura mater. Associated venous hypertension is considered an additional cause for the development of dural arteriovenous shunts, as it may stimulate angiogenic activity and DAVF formation. Other predisposing or triggering factors include pressure changes during air travel, bagatelle trauma, hormonal changes during pregnancy and menopause, systemic hypertension and arteriosclerotic diseases, basic fibroblast growth factor, vascular endothelial growth factor or resistance to activated protein C.

Several classifications of cavernous sinus fistulas have been developed over the years, either based on etiology (spontaneous versus traumatic), the type of the arteriovenous shunt (direct versus

indirect), the type of venous drainage (anterior versus posterior) or hemodynamics (low-flow versus high-flow). The most widely applied classification groups spontaneous lesions into Type A–D, among which Type B–D are DCSFs based on the origin of their arterial supply. More recently, the type of venous drainage pattern and its role in the natural course of the fistulas has been considered. However, none of these attempts to classify DCSFs is unanimously accepted, or can fully satisfy the need for a guide in prognosis, clinical decisionmaking or treatment indications. As the arterial anatomy of these fistulas is of lesser importance in the era of occlusive transvenous treatment, simply dividing them into direct and indirect fistulas may be sufficient for practical purposes.

The signs and symptoms in patients with low-flow DCSFs are in principle similar to those with direct high-flow CCF, but commonly milder and less progressive. They are influenced by size, location of the AV shunt as well as by the type of venous drainage. Clinical presentation in the initial stage can be nonspecific with retro-orbital headaches, mild conjunctival injection or isolated diplopia. Consequently, the disease may be overlooked or is mistaken as endocrine orbitopathy, conjunctivitis or ocular myositis. More advanced stages may present with proptosis, chemosis, retinal hemorrhages or even visual loss. Patients with dominant posterior drainage can present with so-called white-eyed cavernous shunt and may remain undiagnosed for months or even years. Rare differential diagnoses also include orbital tumors or phlegmon. Neglecting a DCSF in the clinical differential diagnoses causes progression of the disease with potentially serious deterioration of the patient’s symptoms and the risk of vision loss. Neurological deficits or intracranial hemorrhage associated with DCSFs are seldom observed (1.5%), despite a relatively frequent occurrence of cortical venous drainage (31%).

Various non-invasive imaging tools are available to detect or rule out a cavernous arteriovenous shunt, including CT, CTA and MRA. Transorbital or transcranial Doppler sonography are useful modalities for screening and confirming an initial clinical diagnosis, or for follow-up. Some clinicians use a pneumotonometer to diagnose DCSFs. For definitive diagnosis and optimal treatment planning, intra-arterial bi-plane DSA remains the gold standard diagnostic test and is considered indispensable in most institutions. Up to the present time, some patients with DCSFs are misdiagnosed, experiencing a frustrating and unpleasant clinical course. Thus, any questionable case should undergo intraarterial superselective angiography as early as possible. The risks of neurological complications associated with cerebral angiography in experienced hands are very low nowadays (1.3% transient, 0.5% permanent complications).

Modern high-resolution imaging including three-dimensional DSA and angiographic computed tomography (ACT) has revolutionized neurovascular imaging providing novel insights into complex anatomic structures such as the CS and its adjacent environment. The often complex arterial supply by numerous dural branches arising from the artery of the foramen rotundum, ascending pharyngeal artery, middle meningeal or accessory meningeal arteries can be visualized in high-resolution cross-sectional images providing a novel perspective as compared to traditional AP and lateral standard angiographic views. The arrangement of efferent CS veins that are obscured or difficult to identify in standard 2D angiograms, such as the inferior petroclival vein, the internal carotid venous plexus or the anterior condylar confluens can readily be depicted, complementing and expanding our existing anatomic understanding. Precise knowledge of venous anatomy and good three-dimensional understanding remains essential for safe and effective endovascular treatment of arteriovenous shunting lesions of the sellar and skull base region.

In addition to endovascular occlusion techniques, therapeutic options of DCSFs encompass conservative management or alternative treatment methods such as manual compression, controlled hypotension and radiosurgery. Although fistula occlusion using TAE with particles can be accomplished in 31%–88%, recanalization has been observed in 25%–100% of cases. Untoward embolization causing neurological deficits with

paresis, aphasia, CN deficits or even intracranial hemorrhage as well as limited long-term durability remain major disadvantages. This has changed the role of TAE into a more adjunctive modality for cases where TVO cannot be utilized or is contraindicated. The reduction of AV shunt flow to increase the efficacy of radiosurgery in intractable cases is another indication. Recent advances in the endovascular armamentarium, such as the introduction of new liquid embolic agents (Onyx), have significantly improved safety and efficacy of TAE of DAVFs. This paradigm shift may also affect future treatment strategies in DCSFs.

Improved knowledge and understanding of the venous anatomy at the CS region and skull base, as well as the development of newer catheterization tools have resulted in the exploration of different venous access routes that can be employed either as alternatives or as complements to achieve definite cure. The establishment of various detachable coil systems contributes to the increase in anatomic occlusion rates up to 100%, while the rate of clinical cure ranges from 52%–100%, with the majority of groups achieving more than 90%. This is superior to results achieved during the early era of endovascular treatment and better than outcomes of TAE series, thus reflecting the level of experience and skill gained in utilizing transvenous occlusion techniques. Current complication rates derived from larger series (n = 613) is 11.6% for transient and 1.8% for permanent deficits (stroke 0.5%).

Key to successful, safe and effective transvenous treatment is the willingness to accept a “multipleroute” approach that eventually enables one to gain access to the CS. Beginning with the simplest and most straightforward IPS approach, only a stepwise increase towards technically more difficult and more aggressive methods, including the transcutaneous SOV approach, is reasonable. Premature utilization of more invasive techniques such as direct puncture of the SOV or CS, as well as combined sur- gical-endovascular techniques should be considered as a last resort with regard to the benign nature of the disease. In suitable cases with mild symptoms, conservative management and appropriate clinical follow-up may be indicated. Manual compression therapy can be applied in compliant patients without contraindications. Controlled hypotension or radiosurgery may be valuable alternatives, especially in patients with difficult anatomical access, contraindications for endovascular treatment or significant comorbidities. Direct surgery for treat-

ment of DCSFs has a minor, complementary role in exceptionally intractable cases, and should be abandoned if at all possible.

To date, hemodynamic characterization of DCSFs is insufficient and mainly based on measurements of flow velocity in the SOV or in ECA branches using percutaneous Doppler ultrasound. Although data on intrasinus flow or pressure are scarce, they may be necessary to fully understand the pathophysiology and natural course of the disease. The characterization of a DCSF as benign, “low-flow” lesions may be too general and insufficient, especially if thrombosis and venous outflow restriction lead to venous hypertension. These

types of fistulas may be more appropriately identified as high-pressure AV shunts.

In conclusion, modern radiologic imaging has remarkably improved diagnostic tests for patients with DCSFs. However, some fistulas are still diagnosed late in the course of the disease or remain unrecognized, leading to frustrating delays in the appropriate therapeutic management. The current status of angiographic equipment, as well as the quality of endovascular tools and devices available today, not only allow a timely and accurate diagnosis, but also demand an early, safe and effective therapeutic intervention whenever angiographically or clinically indicated.