In summary, despite major advances in diagnosis and management, the etiology and pathogenesis of DAVFs and DCSFs remains a matter of controversy and is far from being fully understood. The development of dural arteriovenous shunts must be seen in close relation with thrombotic processes and venous hypertension in the cerebral sinuses. Therefore, the term “venous diseases” appears suitable. To what extend other factors trigger or contribute to the development of dural AV shunts, remains to be investigated by further experimental and clinical studies. Better understanding of mechanisms of molecular pathogenesis in the development of dural AVFs might aid in the establishment of new therapeutic measures for this unique vascular disease.

5.3 Prevalence

Epidemiologic data regarding the overall incidence of both DAVFs as well as of DCSFs are limited. Large autopsy series have found 46 AVMs among 3200 brain tumors (1.4%) (Olivecrona and Riives 1948). Population-based data showed an incidence of 1.84 per 100,000 person years during 1965–1992; the incidence of symptomatic cases was 1.22 per 100,000 person years (Brown et al. 1996). The detection rate in a more recent study was 0.29 per 100,000 adults (Satomi 2008).

According to older literature, DAVF comprise 10%–15% of all intracranial arteriovenous malformations. Most recent authors refer to the work of Newton and Cronvist (1969) who reported on a total of 129 patients among whom 94 had pure pial, 20 had mixed pial-dural and 15 (12%) had pure dural supply. One recent review on natural history of intracranial vascular malformation does not provide new numbers on incidence and prevalence of DAVFs (Brown et al. 2005).

Most patients with DAVFs are more than 40 years of age (82%) and women (71%) (Mironov 1995). Because similar diagnostic and therapeutic measures apply to all dural fistulas, DCSFs and DAVF can be considered as one epidemiologic group of disease. For both, the association with menopause (54%) and puerperium (18%) and lower incidence in men is typical (Newton and Hoyt 1970; Tomsick 1997b; Toya et al. 1981).

In a metaanalysis published by Awad et al. (1990), with an additional 17 cases, the incidence of AVFs in

the CS among all DAVFs was 11.9%, whereas the relationship of “aggressive” to “non-aggressive” cases was 1:6.5 (Table 5.1). In his review of 322 cases, published up to 1920, Sattler (1920, 1930) found a frequency of traumatic vs idiopathic fistulas of 3:1. As elucidated above, this distribution changed towards the end of the twentieth century, and traumatic CSFs were less frequently seen in the western world. This trend is likely due to improved safety standards for traffic vehicles, in particular motorcycles and bicycles. Even large cities in Asia with a high percentage of motorcycle and bicycle riders have a decreased incidence, as recently reported from the Queen Elizabeth Hospital in Hong Kong, where among 80 patients with CSF, 76 (95%) were spontaneous and only 4 (5%) were of traumatic origin (Cheng 2006). Among all CSFs, approximately 10%–15% account for indirect or dural fistulas (Barcio-Salorio et al. 2000) and Type D is by far the most frequent (Tomsick 1997b).

The true prevalence of DCSFs is difficult to assess due to the fact that many patients present with mild symptoms, may undergo spontaneous resolution and are never diagnosed. According to series of several major endovascular centers, Type B–D DCSFs occur five times more frequently than Type A fistulas (Tomsick 1997b).

Another difficulty in obtaining accurate numbers lies in the fact that many DCSFs patients are studied as different groups, either included in DAVFs, CCFs or evaluated as a separate entity. The series of Cognard et al. (1995) contained 205 cases of DAVF of which 33 were DCSF (16%). Klisch et al. (2003) reported on 17 CSFs, including 11 DCSFs. Satomi et al. (2002) reported on 117 cases of benign DAVF (without cortical venous drainage) with DCSF representing the largest group (42.7%). Tsai et al. (2004) investigated 69 patients with DAVFs among whom 30% involved

Table 5.1. Frequency of DCSFs in large series of DAVFs

the CS. Chung et al. (2002) found the CS as the most common (57%) location for the DAVF in 60 patients. Cheng et al. (2003) reported on 27 patients who were assessed based on Cognard’s classification.

Tomsick (1997b) observed Type B–D fistulas in 68% of all spontaneous CSF and discusses the influence of the referring praxis. Table 5.2 provides an overview on published series of spontaneous, traumatic and dural fistulas involving the CS. It reveals that the distribution of various fistula types differs considerably among the groups. While Andoh et al. (1991) reports an incidence of Type D fistulas of 25% and Liu et al. (2001) noted 54%, this type is observed by Vinuela et al. (1984) and Takahasi and Nakano (1980) in 100%.

As discussed above, diagnostic quality of cerebral angiography as major tool for identifying, classifying and understanding etiology and natural history of CSFs has significantly improved over the last 25 years. This must be considered when looking at data in older series using angiographic assessments. Today, high-resolution DSA provides more morphological information than was available in the past and helps to further develop theories and new etiological concepts. Thus, it happened probably not completely by chance that in several recent studies, categorization of DCSFs as groups with different arterial supply (Types B–D) has no longer been applied (see Table 5.2) (Meyers et al. 2002; Cheng et al. 2003; Stiebel-Kalish et al. 2002).

5.3.1

Natural History

DCSFs, in general, are referred to as “benign” or “non-aggressive” fistulas because of their tendency to occlude spontaneously, usually caused by CS thrombosis. Although these spontaneous occlusions seem to occur more frequently than in other AV shunting lesions, they can be accompanied by serious clinical deteriorations. Neither such a course nor a neurological deficit caused by venous ischemia or intracranial hemorrhage, both rare events, can in fact be considered “benign”. But even though such a clinical course may be progressive or in some cases fulminant, terms like “aggressive”, “malignant” or “benign”, widely used by neuroradiologists, appear unsuitable and rather imprecise to characterize a dural arteriovenous shunting lesion. They should be reserved for diseases for which they were defined, i.e. tumors.

As true for prevalence and incidence of DCSFs, exact data on natural history do not exist, which is in part due to the fact that a large number of fistulas is discovered relatively late in their course. Furthermore cases undergoing diagnostic angiography are necessarily affected by the angiographic procedure itself as contrast injection can accelerate thrombosis of the CS and “spontaneous occlusion” (Newton and Hoyt 1970; Seeger et al. 1980; Voigt et al. 1971; Phelps et al. 1982). In some recent series the natural course is additionally influenced by particulate arterial embolization (Satomi et al. 2005; Suh et al. 2005).

The number of reported “spontaneous” occlusions reported in the literature may lie between 11% and 90% (Vinuela et al. 1984; Kupersmith et al. 1988) and is on average 35% according to Tomsick (1997b). When looking at rates of spontaneous occlusion after angiography one has to consider the number of patients reported. For example, frequently quoted, Phelps et al. (1982) observed a “43%” (in Meyers et al. 2002) occlusion rate in 3 out of 7 patients undergoing angiography in a group of 19 patients with atypical signs of a carotid cavernous fistula! This rate is lower in larger series. Only one case in my own material developed a spontaneous occlusion, a patient with an IPS fistula, draining via the CS and causing diplopia (see also Sect. 9.1 and Case illustration XII).

Throughout the process of spontaneous occlusion the progress of thrombosis in the CS and its effects on the venous drainage is uncertain. In many patients regression of ophthalmological symptoms is preceded by an exacerbation of ophthalmologi-

cal symptoms. CS thrombosis, spontaneous or triggered by embolization, may involve the ophthalmic veins and can cause significant deterioration of symptoms due to sudden increase of venous pressure (Seeger et al. 1980). This phenomenon has been described by Sergott et al. (1987) as “paradoxical worsening”. In some cases even vision loss may occur (Sergott et al. 1987; BianchiMarsoli et al. 1996; Suzuki et al. 1989; Knudtzon

1950; Miki et al. 1988). Because it may accentuate intrinsic thrombotic effects on the central retinal vein, an embolization procedure in a patient with worsening of symptoms might be contraindicated (Tomsick 1997b). If increased intraocular pressure occurs, additional anticoagulation may be advisable. Subcutaneous use of low-molecular heparin has improved clinical signs in four patients as reported by Bianchi-Marsoli et al. (1996).

Suh et al. (2005) have retrospectively studied the evolution of fistulas over a mean follow-up period of 23 months. They found that seven (30%) of their patients, angiographically classified as proliferative type (PT), showed chronological progression to the late restrictive type (LRT). Unfortunately, it is not clear from their description how many of these patients underwent treatment by embolization which makes it presumptuous to apply the results to the natural history.

This is similarly true for the study of Satomi et al. (2005) who categorized patients according to their changes in venous drainage but included embolized patients as well.

As for DAVFs, presence of leptomeningeal drainage is a main indicator for risk assessment and deci- sion-making in patients with DCSFs by most authors. The number for cortical or leptomeningeal reflux varies from 10% to 31% (Table 5.3). Nevertheless, the associated risk of intracranial hemorrhage (around 2%) seems, however, relatively low compared to patients with DAVFs in other locations, especially at the tentorial sinus or in the anterior cranial fossa (Agid 2009). For example, none of the DCSF patients in the study in Cognard’s series (Cognard et al. 1995) showed an “aggressive” course and in another large study no case with such behavior was found (Awad et al. 1990). Consequently, cortical or leptomeningeal drainage in DCSFs must not necessarily be seen an indicator of a progressive or “malignant” course or nature, as in DAVFs. Yet today, most operators will probably agree to treat these fistulas as early as possible even though they may pose a lesser risk than direct CCFs with cortical drainage.