Introduction

The aggressive type of intracranial dural arteriovenous fistulae (dAVF) with either direct (non-sinus-type dAVF) or indirect (sinus-type dAVF) retrograde filling of a leptomeningeal vein has a high propensity of ischemia, intracranial hypertension, and cerebral bleeding [8, 44]. If left untreated, the mortality rate is 10.4%, the annual bleeding and the nonhemorrhagic neurological deficit risks are 8.1% and 6.9%, respectively. This leads to an annual event-rate of 15% [44]. Even benign dAVFs bear a risk of 2% to develop cortico-venous reflux with a subsequent risk of complications [33].

Therefore, consensus exists, that aggressive-type dAVFs should be treated [24, 6, 8, 10, 17, 22, 24, 46]. However, dAVFs are rare vascular lesions and large institutional series published so far are seldom [8, 18]. This explains why no clear consensus exists on treatment modality (surgery and/or embolization) and, if surgery is performed, on the best surgical strategy [7, 8, 12, 16, 18, 22, 23, 26, 28, 3235, 37, 40, 44, 46, 48]. After having treated 42 dAVFs microsurgically, we feel encouraged to report our results and to present our surgical strategies.

Materials and methods

Study design

We retrospectively reviewed the medical records of patients with surgically managed dAVFs between 1990 and 2010 either at the department of neurosurgery at the University of Göttingen (2005–2010) or the University of Technology Aachen (1990–2004). The series is a consecutive surgical, but not un-biased series, because in five patients with sinus-type fistulae and four patients with non-sinus-type fistulae embolization was initially attempted without success.

Patients

There were 42 patients (33 male, 9 female, median age of 58.4 years) with angiographically proven dAVFs (Table 1). Sinus fistulae (Fig. 1a and b) were found in 12 cases. The fistulae were located on the sigmoid/transverse sinus (n = 6), the superior sagittal sinus (n = 1), straight sinus (n = 1), cavernous sinus (n = 3), and superior petrous sinus (n = 1). There were 30 patients with non-sinus-type fistulae (Fig. 2a, b and c), with pure drainage into arterialized cortical veins, which were found at the convexity (n = 7), temporal (n = 3), posterior fossa/foramen magnum (n = 4), anterior fossa (n = 7) and tentorial (n = 9) as subsumed in Table 1. According to the angiographic criteria, 12 fistulae had been classified Borden type 2 and 30 fistulae Borden type 3 [4] according to the Borden Classification as described in Table 2.

Table 1 Patient collective
Fig. 1
figure 1

a Preoperative angiographic finding of a sinus-type (sigmoid sinus) fistula. b Postoperative angiography demonstrating complete fistula occlusion after microsurgical sinus packing

Fig. 2
figure 2

a Preoperative angiographic finding of a non-sinus-type (temporal) fistula b: Three-dimensional reconstruction of CT angiography showing a non-sinus-type fistula: the images are suggestive of an occipital fistula with ectasies of arterialized leptomeningeal vein. Intraoperatively, those ectasies were found to be completely within the dural layers with the origin of the leptomeningeal vein in the anterior temporal fossa. c Postoperative angiography proving complete occlusion of the fistula following temporo-dorsal trephination and coagulation of the leptomeningeal vein

Table 2 The Borden classification system

Clinical findings

Clinical findings were epileptic seizures, pareses, aphasia or dysphasia, visual deficits, headache, gait disturbances, tinnitus, loss of consciousness, vertigo, dysmetria, apraxia, bulbous protrusion, and chemosis in 9 (21.4%), 7 (16.7%), 7 (16.7%), 5 (9, 11), 6 (14.3%), 5 (11.9%), 3 (7.1%), 3 (7.1%), 2 (4.8%), 2 (4.8%), 2 (4.8%), 2(4.8%) and 1(2.4%) cases, respectively (Table 3). Sixteen patients (40.5%) suffered from intracerebral or subarachnoid hemorrhage. Multiple findings occurred in 12 patients (28.6%). Fistulae were discovered incidentally in three asymptomatic patients (7.1%)

Table 3 Neurological findings

Surgical strategy

In the non-sinus-type fistulae, the operative strategy consisted of sole interruption of the arterialized leptomeningeal vein, shortly after the dural entry/exit point, either by coagulation and cutting (n = 22) or by clipping (n = 8). No attempt was made to coagulate the feeding meningeal arteries or to excise the arterialized dura. In nine sinus-type fistulae with easily accessible sinus (transverse, sigmoid, superior sagittal sinus), the operative strategy consisted of exposure of the arterialized sinus segment, which then was packed with oxidized cellulose. No attempt was made to cut out the arterialized sinus segment in the last nine cases. In three sinus-type fistulae (one cavernous type and one straight sinus-type fistulae), the operative strategy was to provide access to the venous system for retrograde catheterization and embolization (Table 4). After surgery, angiography was performed in all patients either immediately after the operation or after recovery.

Table 4 Summary of the angiographic findings and treatment strategies

Follow-up investigations

If postoperative angiography proved complete dAVF occlusion, patients were re-investigated yearly for at least 3 years clinically and radiologically using MRI and/or CT angiography and/or angiographic CT. In the absence of abnormal feeding arteries and/or arterialized draining veins and sinus segments, and stable or improving clinical status, no repeated angiography was performed. The follow-up time was 8 weeks to 10 years, with a mean of 4.2 years.

Results

In 39 cases (92.9%), a total elimination of arteriovenous shunting could be achieved by the initial operation. In an extensive torcular/straight sinus and one ethmoidal dAVF, two operations were necessary for complete fistula occlusion. In two cases of sinus cavernous dAVFs, retrograde embolization via the microsurgically opened venous pathway obliterated the fistula: in one patient, the superior ophthalmic vein was exposed microsurgically, in another patient, the initially inaccessible petrous sinus was opened by disconnecting the arterial outflow and, thereby, increasing the intracavernous pressure. In the third patient, with a cavernous sinus fistula, retrograde venous embolization via the petrous sinus and via the superior ophthalmic vein failed, and microsurgical cavernous sinus packing was successfully performed. In one case of a tentorial dAVF, a minimal venous drainage persisted after surgery. In total, the dAVF occlusion rate was 97.6% (Table 4).

Intra- and postoperative complications

Permanent, surgery-associated, neurological deficits were found in 3 (7.1%) out of 42 patients (hyposmia, hemianopsia, and dysmetria in one case each). Transient morbidity occurred in five (11.9%) patients. Pneumatocephalus in one patient and rebleeding in one patient required a second intervention without permanent neurological deficits. The rebleeding occurred in the only case of sinus-type fistulae, in which the arterialized sinus segment was resected. Cerebrospinal fluid (CSF) leakage developed in one patient, which was successfully managed by lumbar CSF drainage. In one case each, a transient paresis of the abducens or oculomotor nerve developed.

Follow-up investigations and outcome

After a mean follow-up period of 4.2 years, no re-opening of the initially occluded fistula, as indicated by the absence of arterialized cortical veins and/or sinus segments on MRI and/or CT, was seen. The minimal venous drainage after incomplete occlusion of a tentorial fistula, which was repeatedly followed by angiographies, remained unchanged during the follow-up period and was not re-treated so far. Fifteen of 16 patients, who experienced a subarachnoid or intracerebral hemorrhage, experienced incomplete recovery. Eight patients (19.0%) with symptoms related to venous congestion showed an improvement of one or more preoperative clinical symptoms. Tinnitus and headache improved in two patients each (4.8%). Five patients (11.9%) were seizure-free after surgery.

Discussion

Dural arteriovenous fistulae are rarer than other intracranial vascular malformations. The development of dAVFs after sinus thrombosis, operation, and trauma involving the sinus had been described. Furthermore, animal experiments indicate that venous hypertension could induce dAVFs possibly mediated by vascular endothelial growth factor [14, 21, 34, 39]. Thus, it is assumed by many that dAVFs are acquired lesions [7].

In 1990, Awad et al. performed a meta-analysis of 377 patients with cranial dAVFs previously presented in the medical literature. They concluded that the pattern of venous drainage is the primary determinant of aggressive behavior [2]. If the flow-through dAVFs empties into leptomeningeal veins either directly or via an arterialized sinus, it can cause venous tortuosity, variceal dilatation of pial veins, and local or regional intraparenchymal venous congestion which has a high propensity to result in nonhemorrhagic neurological deterioration, ischemia, intracranial hypertension, or cerebral bleeding [40]. Patients with aggressive dAVFs have an annual event-rate of 15% [44]. These data are confirmed by Singh et al. who analyzed 73 patients with intracranial bleeding out of 402 patients with dAVFs. They concluded that the bleeding risk, among others, rises with cortical venous drainage [35]. Independently from each other, Borden et al. and Cognard et al. proposed a classification of dAVFs based on the angiographic characteristics of non-aggressive and aggressive fistulae in 1995 [4, 7].

Consensus exists, that aggressive dAVFs should be treated to reduce the risk of bleeding and neurological deficits, but there is an ongoing discussion if microsurgery or endovascular obliteration or a combination of both offers the chance to eliminate the arterial flow in the leptomeningeal-draining vein. The rarity of dAVFs per se, the different angioarchitecture of sinus-type and non-sinus-type fistulae and the different fistula sites explain why no studies exist which compared microsurgery with endovascular therapy.

Non-sinus-type fistulae

In non-sinus-type fistule, the institutional strategy was to interrupt the arterialized leptomeningeal vein, shortly after the dural entry/exit point without attacking the meningeal feeding arteries or excising the arterialized dura. Because these draining veins have inverted flow, the occlusion at their dural origin is not followed by further venous congestion [8, 16, 22, 40, 43, 45]. After occlusion, the draining vein thromboses without significant hemodynamical changes due to downstream veins that already have resumed the draining of functional cerebral areas. We achieved an overall occlusion rate of non-sinus-type fistulae in 97.6%. The permanent surgery-associated morbidity rate for the whole series was 7.1%, and for the non-sinus-type fistulae 4.8% (n = 2). The induced neurological deficits could be considered to be minor. Neuronavigation, using CT angiographic data, proved to be helpful for intraoperative identification of the first segment of the leptomeningeal-draining vein [18]. An increasing number of publications support this concept and report comparable occlusion and morbidity rates [8, 16, 18, 22, 40]. Experiences with the endovascular management of non-sinus-type dAVFs likewise have been reported, but in many series the occlusion rate is lower and the complication rate higher [46]. With transarterial embolization, using particles or N-butyl-2-cyanoacrylate occlusion rates between 28% and 33% could be achieved [40, 43]. Better results were obtained by Agid et al., who achieved angiographic cure in 63.6% [1]. First experiences indicate that transarterial embolization with Onyx has the potential to yield higher occlusion rates. Jiang et al. report an occlusion rate of 68.4% [47]. In the series of Carlson et al. and Nogueira et al., the occlusion rate was even higher with 80% and 83%, respectively [5, 28]. Exceptionally good results were reported by Rossitti, who achieved a 100% occlusion rate [31]. Retrograde transvenous routes are rarely used in non-sinus-type fistulas due to the assumable high risk of rupture of the already fragile leptomeningeal-draining vein [9, 26, 44]. Some combined transarterial and transvenous techniques. In a very recent paper, an occlusion rate of 96% had been reported in 25 patients [19]. However, it remained unclear if true non-sinus-type fistulas were treated in this series because an involvement of the sinus was described for all cases. Complication rates of 9%, occasionally of up to 21%, and even fatalities have been reported in contemporary endovascular series [19, 41, 45, 46, 48]. In consequence, we propose that microsurgical interruption of the draining vein should be the treatment of first choice in all patients with non-sinus-type dAVFs. This especially holds true because the leptomeningeal-draining vein is easily accessible by standard neurosurgical approaches in most instances [25]. The very recent papers of Rossitti and Kirsch et al. indicate that advanced endovascular techniques yield results which are comparable to surgical results. Possibly, our proposal has to be re-evaluated with further up-coming endovascular series. Arterial embolization to mitigate blood loss during surgery is proposed by some [18, 30]. However, in non-sinus-type fistulae we never experienced brisk bleeding during coagulation or clipping of the leptomeningeal vein, and therefore question the usefulness of preoperative arterial embolization.

Sinus-type fistulae

In sinus-type fistulae of sinuses, which are easily accessible by surgery (sigmoid/transverse sinus, superior sagittal sinus), our institutional strategy was to expose the arterialized segment of the sinus, to incise the wall minimally and to pack the sinus with oxidized cellulose until cessation of arterial blood flow. We achieved an overall occlusion rate of 100%, without major morbidity and mortality, and a minor morbidity rate of 10%. These results are in line with the few other surgical series [10, 17]. In these surgical series, sinus packing was achieved with microcoils after sinus puncture. If sinus packing with coils is more advantageous than minimal sinus wall incision and packing with oxidized cellulose will remain a matter of discussion [17]. Possible arguments against sinus packing with coils are the higher costs and the need for intraoperative angiography or patient transfer [29, 30]. In the authors’ opinion, there is rarely the need for microsurgical excision of the diseased sinus segment for control of the dAVF [3, 8, 30]. Furthermore, excision of the arterialized sinus could lead to complications as reported in the literature and witnessed in one case of our series [38]. Van Dijk et al., who only disconnected the cortical venous reflux from the arterialized sinus, proposed a new surgical concept [44]. However, taken the assumed pathophysiology of sinus-type dAVFs, recruitment of new leptomeningeal veins seems possible [14, 39]. Furthermore, the assumption of van Dijk and co-workers that intradural disconnection of the arterialized cortical vein is less risky than sinus packing, which represents a completely extradural operation, could be challenged. In sinus-type fistulae, the retrograde transvenous embolization yields results, which are comparable to surgical sinus packing, with an obliteration rate of 81% to 87.5%, and a permanent morbidity rate of 4% [19, 32]. Transarterial embolization with particles or glue often yields poorer results. Recently, transarterial embolization using Onyx has been proposed for the management of sinus-type fistulae [36]. The reported occlusion rate was 78%. Another endovascular option is conversion of an aggressive into a non-aggressive fistula type by recanalization of the thrombosed sinus, but only single cases have been reported so far [13, 27]. The comparable results of microsurgery and retrograde transvenous embolization do not allow the definition of the superior therapeutic option, but the lower invasiveness favors transvenous embolization as the first treatment option. However, microsurgery should be considered early if transvenous embolization fails.

Cavernous sinus dural arteriovenous fistulae

The cavernous sinus is not easily accessible by microsurgery and sinus packing is not without risk for the cranial nerves [42]. Furthermore, cavernous sinus fistulae could either be bilaterally fed from the carotid arteries or could extend to the contralateral side via the intercavernous sinus, making unilateral sinus packing often ineffective. Thus, cavernous sinus dAVFs are the domain of endovascular interventions [6]. In the rare cases in which the cavernous sinus cannot be reached transvenously, surgery is indicated to provide access to the cavernous sinus via the petrosal vein [15], or the superior ophthalmic vein [49] for retrograde catheterization. In one case, we disconnected the arterialized leptomeningeal-draining vein and thereby successfully increased the intracavernous pressure, which “opened” the blocked petrous sinus for successful retrograde embolization, which is a strategy that has not been reported so far. Very rarely, embolization proved to be not possible, as witnessed in one patient, and microsurgical sinus packing remains the only treatment option.

Limitations of the study

In many surgical and radiosurgical and some endovascular series, angiographic follow-up investigations in addition to early angiography after surgery were performed only sporadically or within less than 4 months [5, 16, 18, 20, 22, 31, 42]. Our study shares this limitation. From the scientific standpoint, repetitive follow-up angiography would be of interest to prove sustained occlusion of the fistula and to exclude distant fistula development. On the other hand, surgical series and recent endovascular series of transvenous embolization indicate that re-opening of the fistula is very rare after complete occlusion, which probably justifies that non-invasive imaging modalities such as MRI and CT angiography or, very recently, intravenous angiographic CT are used for follow-up imaging [12, 36, 44, 47].

Other treatment options

In 2005, Koebbe et al. described 18 cases of dAVF treatment either with gamma knife radiosurgery alone or in combination with embolization [20]. They achieved a total obliteration rate of 83.3%. Friedman et al described their experiences with combined stereotactic radiosurgery and transarterial embolization for dAVFs of the transverse and sigmoid sinus. Although symptoms resolved or improved in 96% of the patients, 59% of the lesions were observed to persist to some degree on follow-up angiograms [11]. Apart of the comparatively low occlusion rate, it has to be taken into account that it takes months to years for thrombosis to occur, which is not acceptable if dealing with dAVFs of the aggressive type. At least if dealing with aggressive dAVFs, surgery and embolization remain the mainstays of the therapeutic management.

Conclusion

Our surgical strategy was to focus on the arterialized leptomeningeal vein in the non-sinus-type fistulae and on the arterialized sinus segment in the sinus-type fistulae. With this strategy, the operative morbidity is low and the dAVF occlusion rate is high in both non-sinus and sinus-type fistulas. In the light of these results, we propose that microsurgery should be considered early in the management of aggressive cranial dAVFs. The only exceptions are selected cases of cavernous sinus dAVFs in which the role of surgery is not the elimination of the fistula, but the establishment of an access to the venous system for retrograde embolization.