Surgical Management of Cranial Dural Arteriovenous Fistulas

Pathogenesis

The etiologic factors and mechanisms involved in the pathogenesis of DAVFs are incompletely understood. DAVFs are associated with several conditions including head injury, previous craniotomy, and dural venous sinus thrombosis, suggesting that they are acquired rather than congenital lesions. ^, According to the most widely accepted theory, DAVFs are formed as a consequence of thrombosis and subsequent recanalization of dural venous sinuses. ^, Venous hypertension is thought to play an important role in this process. Animal studies have shown that venous hypertension, even in the absence of sinus thrombosis, can elicit the formation of DAVFs. Whether sinus thrombosis is in fact the initial event in the genesis of DAVFs is controversial. Only a small number of patients with sinus thrombosis go on to develop DAVFs and not all DAVFs are associated with sinus thrombosis. ^, What follows sinus thrombosis and venous hypertension is also controversial. Two hypotheses have been proposed. The first suggests that DAVFs arise from the opening up of preexisting microscopic vascular channels within the dura mater. These preexisting channels are thought to open up or enlarge as a result of venous hypertension secondary to sinus thrombosis. The second hypothesis suggests that DAVFs result from the formation of new vascular channels in the dura, a process stimulated and regulated by angiogenic factors. To support this, surgical DAVF specimens have been shown to contain basic fibroblastic growth factor and vascular endothelial growth factor, which were absent in control specimens. ^, Angiogenic factors may originate either directly as part of the inflammatory process that occurs during organization and recanalization of a thrombosed sinus or indirectly as a result of cerebral ischemia secondary to venous hypertension. If the angiogenic theory is true, antiangiogenic agents may provide an adjuvant therapy for patients with untreatable DAVFs. Many cranial DAVFs ultimately undergo spontaneous resolution. The exact mechanism for this is unknown but it is thought to result from progressive thrombosis of the involved dural sinus. This is paradoxical in that the same pathologic process underlying the abnormality is thought to lead to the process of resolution of the abnormality. In some cases, however, spontaneous resolution has occurred despite sinus patency.

Classification

Ideally, a classification system should predict the clinical behavior of a lesion and aid in therapeutic decision-making. Several classification schemes have been devised for DAVFs. The most commonly used are those of Borden and colleagues (Borden classification) and Cognard and colleagues (Cognard classification) ( Table 54.1 and Fig. 54.1 ). Both are based on the pattern of venous drainage of the lesion, the factor that best predicts the clinical presentation and natural history of DAVFs. The Cognard classification, which is a modification of the classification of Djindjian and colleagues, divides cranial DAVFs into five types based on the presence or absence of CVR, sinusal drainage, and direction of flow in the involved dural sinus.

The Borden classification is a more simplified scheme with three types based on the presence or absence of CVR and sinusal drainage without taking into account the direction of flow in the venous sinus. In a study by the University of Toronto Brain Vascular Malformation Study Group, Davies and colleagues validated the classification systems of Borden and of Cognard with respect to clinical presentation. Aggressive presentation (i.e., intracranial hemorrhage, focal neurologic deficit, or death) occurred in 2% of the Borden classification type I, 39% of type II, and 79% of type III cranial DAVFs. Similar correlation was found between the Cognard classification and clinical presentation ( Table 54.2 ). Subsequent studies have demonstrated that the pattern of venous drainage and in particular the presence or absence of CVR also correlate with the natural history of DAVFs after the initial presentation. ^,

DAVFs were previously classified according to their anatomic location. Anterior cranial fossa and tentorial lesions were associated with a higher risk of aggressive clinical behavior than lesions in other locations. ^, However, it was later shown that the poor prognosis of lesions in these “dangerous” locations was purely a function of their pattern of venous drainage. ^, It is the presence of CVR, not the anatomic location per se, that leads to aggressive clinical behavior. ^, Hemodynamically, DAVFs may be classified into high- or low-flow fistulas. The classification of Barrow and colleagues is described later in the section on carotid cavernous fistulas. The classification of pediatric DAVFs is not discussed in this chapter.

Clinical Presentation

Other than pulsatile tinnitus, the symptoms of DAVFs are related to venous hypertension. ^, This may lead to venous congestion, cerebral edema, cerebral infarction, and intracranial hemorrhage (ICH). The clinical features of cranial DAVFs are shown in Table 54.3 . ICH, nonhemorrhagic neurologic deficit (NHND), and death are considered aggressive. , The most significant risk factor for aggressive clinical presentation is the presence of CVR. ^{, ,} Therefore type II and III lesions of the Borden classification are frequently associated with aggressive clinical behavior, whereas type I lesions are rarely aggressive (see Table 54.2 ). ICH associated with DAVFs typically results from rupture of an arterialized cortical vein and is usually intraparenchymal, but it may also present in the subarachnoid, subdural, and intraventricular spaces ( Fig. 54.2 ). NHND is caused by cerebral ischemia secondary to venous congestion. This category does not include ophthalmoplegia secondary to cranial nerve dysfunction. The neurologic deficit may be focal or global.

Benign symptoms include pulsatile tinnitus and orbital symptoms. Pulsatile tinnitus is produced by turbulent flow in the diseased dural sinus. Objective bruit is heard on auscultation in 40% of patients with tinnitus. Ophthalmologic symptoms and signs are most commonly seen with cavernous sinus DAVFs, but they can also occur with lesions in other locations if the venous drainage involves the cavernous sinus and ophthalmic veins. The symptoms are caused by venous congestion. ^, The ophthalmologic symptoms may be progressive and disabling and may even lead to blindness; therefore they may not be considered benign by the patient.

Radiographic Evaluation

Computed Tomography and Magnetic Resonance

In cases of DAVF without CVR, the results of computed tomography (CT) and magnetic resonance imaging (MRI) of the brain parenchyma are typically normal. However, MRI and magnetic resonance angiography (MRA) may show stenosis or occlusion of the dural sinuses. Hydrocephalus may be seen in any DAVF that causes venous hypertension in communication with the superior sagittal sinus. ^, In cases of DAVF with CVR, CT and MRI of the brain may show ICH, engorged pial vessels, and diffuse white matter edema secondary to venous congestion. The ICH is usually intraparenchymal but may also have subdural, subarachnoid, or intraventricular components (see Fig. 54.2 ). The pattern of hemorrhage is not specific to DAVFs, and a high index of suspicion is required. Dilated pial vessels and white matter edema are more likely to be seen on MRI than on CT. On T2-weighted MRI, dilated pial vessels are seen as flow voids on the surface of the brain, and diffuse white matter edema is seen as T2 hyperintensity or FLAIR signal in the cerebral or cerebellar hemispheres, brain stem, or spinal cord ( Fig. 54.3 ). Recent reports have demonstrated improvements in the ability of MRA to diagnose DAVFs and detect CVR. However, these results have been reported in small selected series and require further study. At present, with negative CT, MRI, or MRA, the diagnosis of DAVF cannot be excluded. If it is clinically suspected, catheter angiography is required to confirm or exclude the presence of a DAVF.

Intra-Arterial Catheter Angiography

This is the gold standard method for the diagnosis and evaluation of DAVFs. Their characteristic angiographic feature is premature visualization of intracranial veins or venous sinuses during the arterial phase (see Fig. 54.3C ; Fig. 54.4A ). This is caused by the shunting of arterial blood into the venous system through the fistula. To obtain the necessary information regarding the arterial supply and venous drainage of the DAVF and the venous drainage of the brain, imaging must start early in the arterial phase and be carried through the venous phase. The study should include injections of both internal carotid arteries (ICAs), both external carotid arteries (ECAs), and both vertebral arteries. This is because a single DAVF may have multiple feeding arteries (see Figs. 54.3 and 54.4 ) and also because 8% of patients have multiple DAVFs. Detailed knowledge of dural arterial anatomy is essential in the angiographic evaluation of DAVFs. Meningeal arteries that are invisible or difficult to see on normal angiograms may be dilated and clearly visible when they are supplying a DAVF. For example, the tentorial branch of the meningohypophyseal trunk of the ICA (the artery of Bernasconi and Cassinari) or the meningeal branch of the posterior cerebral artery (the artery of Davidoff and Schecter) may be dilated and easily seen on the angiograms (see Fig. 54.3C ).

The nidus of a DAVF is the site of arteriovenous shunting and refers to that part of the dura where all feeding arteries and the origins of venous draining channels converge. The best views of the nidus are often obtained during the early arterial phase of the angiogram and by the selective injections of arterial feeding branches. Images obtained when the proximal arterial feeders are injected, particularly in the later arterial phase or the venous phase, are often obscured by engorged feeding arteries and draining veins.

Assessment of the venous drainage pattern of DAVFs is extremely important, as this factor determines the natural history of the lesion and aids in selecting the most appropriate management strategy. The presence or absence of CVR and venous sinus occlusion, the direction of flow in the venous sinuses, and the venous drainage pattern of the brain must be determined. The exact site of CVR must be determined to allow treatment planning. At angiography, a delayed contrast washout time is compatible with venous congestion in the region where it is observed. Focal areas of delayed venous drainage in the brain correspond to the site of CVR. In some cases, tortuous, dilated pial veins may be seen that develop as a result of venous hypertension (see Fig. 54.3F ). Willinsky and colleagues have described this finding as the pseudophlebitic pattern, which is a sign of venous congestion of the brain and may be associated with an aggressive natural history.

Natural History

Knowledge of the DAVF risk factors (discussed previously) and related natural history is essential for guiding patient management. The results and complications of available treatment strategies must be compared with the outcome of the natural history of the disease. Of all risk factors, CVR is the most significant predictive variable for an aggressive natural history. ^,

Indications for Treatment

Decisions regarding the treatment of DAVFs should be guided by an understanding of the natural history. Therefore DAVFs with CVR should be treated to eliminate the risk of hemorrhage and neurologic deficit. The aim of treatment in these cases is the elimination of CVR or complete cure if possible. Lesions without CVR do not require treatment unless they are associated with intolerable symptoms such as tinnitus, bruit, and ophthalmologic symptoms including visual deterioration and pain. For these lesions, the aim of treatment is not cure but rather symptom control.

Management Options

Several options are available for the management of cranial DAVFs. In many cases, a combination of methods may be required. More common types of DAVFs—such as carotid-cavernous, transverse-sigmoid sinus, superior sagittal sinus, and infratentorial—are often amenable to endovascular therapy. However, several types of DAVF may have anatomic restrictions that favor open surgical treatment. These can include straight sinus, tentorial, and ethmoidal DAVFs.

Endovascular treatment of ethmoidal DAVFs has not been widely performed for several reasons. First, selective catheterization of the ophthalmic and ethmoidal arteries is difficult due to their small caliber and tortuous course. Second and most importantly, embolization of the central retinal artery, also a distal branch of the ophthalmic artery, can complicate the procedure and devastate a patient’s vision. Third, unlike embolization of other DAVFs whose blood supply originates from the ECA, embolization of ethmoidal DAVFs through an ICA branch carries the risk of embolic agents refluxing into the cerebral circulation. These obstacles have deterred most endovascular surgeons from treating ethmoidal DAVFs. The results that have been published suggest that endovascular management of ethmoidal DAVFs poses a small but clinically significant risk to vision, is rarely effective in curing the fistula, and does not eliminate the need for surgery.

Galenic, straight sinus, and tentorial DAVFs are difficult to cure with endovascular therapy. Their arterial supply is extensive, involving meningeal arteries from the ICA and vertebral artery that are difficult to cannulate and riskier to embolize than feeders of the ECA. Transvenous navigation to deeper locations around the tentorium is difficult. More importantly, these DAVFs can often drain exclusively to subarachnoid veins rather than to their associated sinus (Borden type III), which prevents transvenous access. Therefore the management of these DAVFs, unlike most other DAVFs, may require microsurgical interruption. ^,