Propofol (Diprivan) is a short-acting anesthetic agent with minimal side effects commonly used for induction or maintenance of general anesthesia and for sedation of ICU patients. It is very feasible to induce short periods of controlled hypotension.

Recently, the same group reported on the longterm outcome in 14 patients with DAVFs in different locations (de Miquel et al. 2005). Five patients (36%) had very low-flow DAVFs and four (28.5%) resolved completely. In one patient, the procedure could not be completed due to hemodynamic instability of the patient. Out of three patients (21.4%) with low AV shunt flow, two (14.3%) occluded and one (7%) significantly improved. In five cases (35.7%) where flow was considered medium, two (14.3%) occluded, one (7%) improved and one remained unchanged. In one high-flow fistula, the flow changed only slightly and the patient had to be treated by coil occlusion. The authors concluded that controlled hypotension is a good adjuvant tool in low-flow DAVFs and may lead to long-term occlusions.

The Case Report (Fig. 9.3) shown here depicted that induced hypotension can be effective, and thus should be considered in cases with small or residual AV shunting. The pathophysiological mechanism is still not fully understood and the role of venous pressure in the pathogenesis of DAVF is discussed in more detail in Chap. 5. Pressure changes, atmospheric or systemic blood pressure, especially when combined with intraarterial contrast application, may possibly promote thrombosis and occlusion in DCSFs, particularly in low-flow cases.

Kupersmith et al. (1988) reported two cases where occlusion of DCSF occurred following air travel. One patient developed choroidal detachment and 3rd nerve palsy; the angiogram performed the next day showed complete thrombosis of the DCSF, ipsilateral CS and SOV. Another patient demonstrated complete resolution of the bruit and bilateral 6th nerve paresis after a plane flight confirming angiographic occlusion the next day.

It is noteworthy that the role of lowered systemic blood pressure has been considered a causative factor in cases that developed “spontaneous” occlusion following angiographic procedures performed under general anesthesia (Potter 1954;

Parsons et al. 1954). Echols and Jackson (1959) observed occasional success with bilateral carotid compression under hypothermia. They reported a 25-year-old patient with a traumatic CCF who underwent hypothermia with a body temperature reaching 85.5˚F, leading to unattainable blood

pressure and peripheral pulsation for 45 min. During this period, the bruit could not be heard, but returned after 1 h, although to a lesser degree. Potter (1954), who observed a case of spontaneous cure after a period of severe hemorrhage and syncope, also discussed the role of hypotension for occlusion of CCFs.

Due to the general risks of adverse reactions associated with anesthetic drugs, the use of controlled hypotension under general anesthesia as a single means for therapeutic management is probably not justified. It appears, however, reasonable as an adjuvant measure in patients undergoing endovascular treatment, excluding those with cerebrovascular, renal or cardiac insufficiency.

9.4 Radiotherapy

Irradiation for treatment of a dural cavernous sinus fistula was introduced by Barcia-Salorio et al. in 1979 in a 65-year old patient suffering from chemosis, exophthalmos and diplopia. After stereotactic radiosurgery using a single cobalt source delivering 40 Gy, the patient was reported to be symptom free 2 months later. In 1982, Bitoh et al. (1982), successfully treated two patients with 60Co (Cobalt) using 32 Gy and 30 Gy achieving complete occlusion, documented by angiography. Radiotherapy of DCSFs has been advocated ever since as an alternate treatment option for DCSFs by several groups (Bitoh et al. 1982; Soderman et al. 2006; Hirai et al. 1998; Yasunaga et al. 1987; Yamada et al. 1984; Pierot et al. 1992; Mizuno et al. 1989; Hasuo et al. 1996; Guo et al. 1998; Moriki et al. 1993; Onizuka et al. 2003; Pollock et al. 1999). Conventional fractionated irradiation via linear accelerator (Barcia-Salorio et al. 2000; Hirai et al. 1998; Yasunaga et al. 1987), and more recently, a 60Co source (Bitoh et al. 1982; Yamada et al. 1984) via gamma knife radiosurgery (Guo et al. 1998; Moriki et al. 1993; Pollock et al. 1999; Link et al. 1996) have been suggested.

In some series, radiotherapy has been combined with preor post-procedure embolization (Hirai et al. 1998; Pierot et al. 1992; Hasuo et al. 1996; Pollock et al. 1999; Link et al. 1996).

In 1982, Barcia-Salorio et al. described gamma knife radiosurgery in four patients as a “bloodless

operation without anesthesia, with no surgical risk that can be applied independent of age and general condition of the patient”.

Twelve years later, the same group (BarciaSalorio et al. 1994a), reported 25 patients treated with a modified 60Co therapy, applying an estimated dose of 30–40 Gy. In all, 22 patients had DCSFs, of which 20 (90%) could be completely obliterated during an average post-treatment period of 7.5 months (2–20). In Type B fistulas, a 100% occlusion rate was achieved, compared to 75% and 86% in Type C and D fistulas, respectively. In two cases, retreatment was required to achieve complete occlusion. The same group later reported an improved overall occlusion rate in DCSFs of 91.6% (Type B: 100%, Type C: 90%) in 24 patients, after a mean follow-up interval of 7.2 months (Barcia-Salorio et al. 2000). There were no post-irradiation injuries to the optical or ocular motor nerves. The occlusive effect is explained by early endothelial swelling that may evolve to partial or complete thrombosis associated with basal membrane rupture, necrosis, interstitial exudates and leukocyte invasion. This leads to fi- broblastic and endothelial cell proliferation resulting in intimal hyperplasia that can be observed after application of doses between 30–40 Gy (Joanes et al. 1991). Hyaline degeneration may occur after 60 Gy (Joanes et al. 1998).

Link et al. (1996) reported on results of gamma knife therapy in 29 patients with DAVFs, among which 10 were located at the CS. In 17 (59%) of these patients who had either pial or cortical drainage, additional transarterial embolization within 48 h after radiotherapy was performed. A reduction of the AV shunting by transarterial embolization may decrease the risk of bleeding in the latency period (post-treatment period) in these high-risk patients. By using this combined approach, complete occlusion was accomplished in 75% and a partial occlusion in 29.5% of cases. One of the patients with DCSF developed a mild transient expressive aphasia after the embolization.

Hasuo et al. (1996) reported on nine patients with type D fistulas, who were treated with 30 Gy after particulate embolization, achieving immediate improvement and complete resolution of the symptoms over 4–19 months. The authors recommend this treatment as thetherapy of choice in type D fistulas with only mild symptoms. Guo et al. (1998) recently published a series of 18 patients with DCSFs (B: n = 1, C: n = 7, D: n = 10), who were treated primarily with gamma knife surgery (22–38 Gy). Target levels were

kept at 50%–90% isodose, while neural structures such as the optic nerve were kept at 8 Gy. Complete occlusion was achieved in 15 patients (83%). The remaining three showed partial obliteration with no complications or worsening of symptoms observed over a period of 27 months. The authors concluded that gamma knife treatment is a feasible and safe alternative for patients with DCSFs.

Hirai et al. (1998) reported the use of multifractionated radiation, performed via a linear accelerator administering a total dose of 30 Gy over 15 sessions (2 Gy per session) in all but one patient, who received 40 Gy over 20 sessions. In 12 patients undergoing irradiation alone, cure was achieved in 75% of cases. Two cases with fast-flow fistulas showed no change; one demonstrated occlusion. In all, 14 patients underwent TAE or TVO. In six of these patients, endovascular treatment was combined with irradiation, achieving cure in 12 patients (86%). Two patients with fast-flow type fistulas demonstrated either no change or improvement of symptoms despite undergoing combined treatment. One patient suffered from ischemic stroke after TAE and recovered within 1 year, while another presented with transient epilation following TVO.

Pollock et al. (1999) treated 20 DCSFs with either radiosurgery alone (n = 7), or radiosurgery followed by particulate transarterial embolization (n = 13), achieving improvement of symptoms in 19 (95%), and total angiographic occlusion in 87%. Two patients who did not show initial angiographic occlusion experienced recurrence of the symptoms and underwent repeated TAEs. One patient underwent TVO. None of the patients who demonstrated angiographic occlusion (14/15) showed recurrent symptoms. Three patients (15%) in this group experienced complications: one patient developed a stroke during stereotactic angiography, one developed venous ischemia and another a permanent 6th nerve palsy due to acute CS thrombosis as a consequence of the embolization procedure.

Soderman et al. (2006) recently reported on 53 patients with 58 DAVFs treated with gamma knife surgery achieving total occlusion in 68% (angiographically proven obliteration) and significant flow reduction in 24%. Five patients (9%) with DCSFs were included receiving a minimal dose of either 10 Gy or 12 Gy resulting in complete obliteration (Case Illustration XIV). One patient (18.8%) with preexisting 6th nerve palsy received a maximal dose of 50 Gy and developed transient worsening of his symptoms. The entire series also contained a case

of late reaction (10 years) with hemorrhage (1.9%), one patient with focal alopecia and two angiography related minor complications.

It is emphasized that DAVFs with cortical venous drainage may not respond to radiation, which is a disadvantage, as these patients have an increased risk of intracranial hemorrhage during the post-treatment period. DCSFs in general may have a lower risk of intracranial hemorrhage than DAVFs in other locations; neurological deficits due to venous congestion may develop during this latency period. Some investigators see this as a major downside of radiosurgery and favor embolization therapy instead (Cognard et al. 2008). It should also be mentioned that in some cases of DCSFs, the time to occlusion is relatively short compared to the time required for obliteration of cerebral AVMs (Heros 2006). In one of the largest series to date, 100% of Type B fistulas closed after a mean period of 5.9 months, 75% of Type C fistulas closed after 12.6 months and 85.7% of Type D fistulas closed after 8.16 months (Barcia-Salorio et al. 1994a). Soderman et al. (2008) reported the same phenomenon, assuming that the location within the dura mater and the usually narrow vessels may play a role in this relatively faster obliteration.

Although complications directly related to radiosurgery of DCSFs are rare, side effects may occur. Lau et al. (2006) reported on paradoxical worsening in a patient with DCSF undergoing radiotherapy who developed signs of SOV thrombosis and central retinal vein occlusion. As the radiation target includes not only the dura mater, but also the venous CS compartments, progressive obliteration in an unpredictable manner might be of general concern. As stressed for other indirect treatment modalities, a benign DCSF may transform into a more aggressive lesion, exposing the patient to increased risk of neurological deficit or intracranial hemorrhage and death. Such an untoward change may occur during the relatively long follow-up period, especially when radiosurgery is used as a single treatment option. To monitor changes in direction of the SOV flow, intermittent imaging follow-up using Doppler ultrasound may be recommended (Chiou et al. 1998).

Definite occlusion of the AV shunt should be affirmed by intra-arterial DSA rather than MRA, as was recently reported by Chiou et al. (1998), who successfully treated four patients, one with cortical venous drainage. Small remaining fistulous compartments are not detectable using MR angiogra-

phy. Although stated otherwise by some investigators relying on symmetric diameters of the SOVs or appearance of the orbital fat tissue (Struffert et al. 2007), MRA was completely unreliable to rule out small AV shunts, especially when draining posteriorly in several of my own cases.

Non-invasive imaging follow-up, if at all, should rather be used for patients undergoing TVO, in whom complete occlusion is mostly achieved at the end of the procedure or will occur within days or weeks post procedure in the majority of cases.

Although undue effects of radiation on the optic or oculomotor nerves have not been reported in patients with DCSFs, they may occur at a rate of 6.3% following radiosurgery of CS tumors (Tishler et al. 1993). A dose limit of 8–10 Gy appears to be accepted for the optic nerve and the brain stem (Guo et al. 1998; Tishler et al. 1993; Barcia-Salorio et al. 1994b), while such limits for intracavernous cranial nerves remain to be determined (Barcia-Salorio et al. 2000).

In summary, radiotherapy of DCSF represents an effective alternative (up to 90%) (Barcia-Salo- rio et al. 2000) and should be considered a valuable complement of the therapeutic spectrum for DCSFs. In selected cases in which endovascular means remain unsuccessful, ineffective, contraindicated or considered too risky, irradiation using a linear accelerator or gamma knife should be considered. In elderly patients with comorbidities, lengthy endovascular procedures under general anesthesia could be avoided, especially when they have already failed in previous sessions. Radiosurgery should focus primarily on small, low-flow shunts and, if possible, be combined with transarterial embolization or manual compression. The efficacy in high-flow lesions or direct CCF appears questionable (Barcia-Salorio et al. 2000). Newer data on the natural history seem to indicate that hemorrhage rates for DAVFs with cortical drainage may be lower than previously expected (Soderman et al. 2008). Nevertheless, radiotherapy should probably not be considered in those cases, or when neurological manifestation is already evident.

Although not seen by the author, so-called “intractable” cavernous sinus fistulas have been reported by others (Link et al. 1996; Guerro et al. 2006). Due to continuous advancement of endovascular tools and devices, improved visualization techniques and enhanced anatomic understanding, their number will also remain relatively small in the future.

9.5 Surgery

Direct surgical treatment of CSFs was the primary treatment modality in the pre-endovascular era; however, it was associated with significant morbidity and mortality (Dandy 1937; Sattler 1930; Locke 1924; Hamby and Gardner 1933). Ligation of the ophthalmic veins was successfully performed by Lansdown in 1874 (Landsdown 1875) and repeated by others (Sattler 1880; Locke 1924) during subsequent years. Locke (1924) reported a 68.4% cure or improvement rate with this treatment, with a mortality of 5.3%. About 100 years after Ashley Cooper performed the first CCA ligation, De Schweinitz and Holloway reviewed 114 patients communicated in the literature to that date, and found a good outcome in 56%, no improvement in 17.5% and a recurrence rate of 20%. The overall mortality rate was 11.7%. Locke (1924) documented cure or improvement in 61.9% of the patients undergoing carotid ligations for treatment of pulsating exophthalmus and a mortality of 14.3%. Because of such high recurrence rates and poor, or even fatal outcomes, ligation of intracranial carotid arteries was proposed by Hamby and Gardner in 1933.

The introduction of microsurgical techniques enables direct surgical approaches to the CS for obliteration of direct and indirect CSFs (Parkinson 1965, 1987; Dolenc 1990; Isamat et al. 1986; Francis et al. 1995; Vinuela et al. 1984). Isamat et al. (1986) described four patients (three traumatic, one spontaneous), whose CS was exposed via pterional craniotomy allowing the introduction of muscle fragments and/or fibrin sealant. The same authors reported on seven additional patients treated by transmural injection of fibrin sealant into the CS achieving complete occlusion in all, preservation of the ICA in 4/4 cases and observing one transient postoperative hemiparesis (14.2%) (Isamat et al. 2000). It is emphasized that control of the hemodynamic condition by placing a temporary clip on the supraclinoid carotid and transient occlusion of the cervical ICA using manual compression or a rubber band can be a useful adjunct.

With the onset of transarterial embolization techniques, several investigators pioneered their use of direct and indirect CSFs (Halbach et al. 1987;

Kerber et al. 1979; Manelfe and Berenstein 1980;

Bank et al. 1978; Picard et al. 1987; Chermet et al. 1977). Particularly since Serbinenko’s introduc-

tion of balloons for occlusion of direct CCFs in 1974 (Serbinenko 1974), direct surgical obliteration of CSFs has been increasingly abandoned.

Despite the establishment of EVT as a primary treatment modality today for direct fistulas and indirect fistulas, open surgery of the CS is still considered useful in certain cases (Isamat et al. 2000; Tu et al. 1997; Day and Fukushima 1997; van Loveren et al. 1991).

In the Tu et al. (1997) series, 19/78 (24.3%) patients with type A–D CSFs who initially underwent EVT were definitively treated by subsequent microneurosurgery. The authors reported an occlusion rate of 100% and an ICA patency rate of 94%. Among these 19 patients, three (15.8%) had a DCSF in which either the small fistula size or multiple ECA branches were reported as reasons for failure of endovascular techniques. Interestingly enough, no transvenous techniques were attempted in any of these patients. Eight patients (42%) experienced transient 3rd nerve palsy, while in one patient (5.2%) permanent 6th nerve palsy was noted.

Van Loeveren et al. (1991) performed direct surgery in eight patients with Type A fistulas between 1979 and 1996, achieving complete occlusion in 75% and reduction of the fistulas in 25%. There were immediate postoperative CN deficits in five patients (63%), among whom two were permanent (one patient was lost to long-term FU).

Day et al. (1997) published a series of nine Type D fistulas treated by a combined extra-intradural approach to the CS and consecutive obliteration of the fistulous communication. In this group, endovascular treatment had been attempted using TAE (de Keizer 2003) and transvenous approaches (Phelps et al. 1982), but failed. It is stated that Type D fistulas represented a “challenge” for endovascular treatment. This may be true for transarterial embolization, but is not for transvenous occlusion techniques using all available routes. It remains unclear to what extent endovascular treatment options were fully explored in some of the cases (Illustrative Case 1). While resolution of the symptoms and angiographic occlusion were complete in 100% of patients, transient diplopia and trigeminal hypoesthesia developed in each patient, resolving over 6 months. In addition, there was a high procedure-related morbidity of 22% [transient hemiparesis in one (11%) and permanent hemiparesis in another (11%)]. The success of this method, which was originally described by Mullan (1979), relies on superb knowledge of the cavernous triangles while “sensing” how to avoid