Article Text
PDF
Abstract
Objective Cribriform plate dural arteriovenous fistula (dAVF) is a rare pathology, for which the treatment of choice used to be neurosurgery. Technological advances in micro-catheters and embolic agents permitted new endovascular alternatives.
Methods We included all patients treated endovascularly for a cribriform plate dAVF between 2008 and 2013. We retrospectively analysed data focusing on the type of treatment chosen.
Results Ten patients were treated by endovascular approach, with a need for an additional surgical exclusion of the fistula in two cases. Thirteen embolisation sessions were done. Embolisation of the fistula through an ethmoidal artery was the technique of choice; the catheterism of the ophthalmic artery was impossible in two cases and the embolic agent did not penetrate in four cases. The embolisation through the middle meningeal artery was successful in one case but the tortuosity of this artery prevented good penetration of the embolic agent. Venous approach was successful in all cases but was limited to fistulas with superficial venous drainage.
Conclusions Endovascular treatment of cribriform plate dAVF is safe and effective. The embolisation through ethmoidal artery is the method of choice. Branches of the middle meningeal artery are tortuous and prevent the penetration of embolic agent. Venous approach is effective but is limited to cases of failure of the arterial approach.
- Fistula
- Hemorrhage
- Intervention
- Liquid Embolic Material
- Vascular Malformation
Statistics from Altmetric.com
Introduction
Cribriform plate dural arteriovenous fistulas (dAVFs) are rare. They receive arterial supply essentially from ethmoidal branches of ophthalmic arteries.1 ,2 Considering the high risk of bleeding or rebleeding of dAVFs in this location,3 there is no doubt about the indication of treatment. The method of choice is the surgical excision of the fistula because of its superficial location, far from eloquent areas and cranial nerves, and its easy exposure by classic fronto-temporal craniotomy.4 There are few reports in the literature regarding endovascular treatment of cribriform plate dAVFs,3 ,5–11 even though technological advances in micro-catheters and embolic agents recently allowed the proposal of new alternatives using the endovascular approach. We present our experience in the endovascular treatment of this subtype of dAVF.
Materials and methods
Demographic data
We have maintained an ongoing prospective registry of patients treated in our centre for dAVF. From 2008 to 2013, 10 patients with dAVF of the anterior cranial fossa were treated by the endovascular approach in our institution. Treatment strategy was discussed in a multi-disciplinary manner, including a neurosurgeon and an interventional neuroradiologist. Demographic data recorded for each patient included age, sex, initial clinical presentation and discovery mode of the dAVF.
Angiographic characteristics of dAVF
Cerebral MRI and selective six-vessel cerebral digital subtraction angiography (DSA) were performed for all patients. Characteristics (location and type) of the fistula were meticulously recorded as well as anatomy of feeding arteries (number, origin) and draining veins.
According to the classification of Lariboisière,12 we classified dAVF according to the venous drainage and venous reflux.
Endovascular procedures
Endovascular procedures were performed under general anaesthesia and systemic heparinisation. After diagnostic angiography was performed, the appropriate working angles for catheterisation of the arterial supply of the dAVF were recorded. Treatment was made by embolisation of the fistulous point and of the venous collector duct by cyanoacrylate synthetic glue (Glubran, GEM, Viareggio, Italy) or by Onyx (ev3 Neurovascular, Irvine, California, USA).
Postoperative follow-up
Follow-up started at the time of treatment and finished with the last visit or angiography. Angiographic follow-up after complete exclusion of the dAVF consisted of 6-month DSA. A complete ophthalmologic examination was done between 6 and 12 weeks after the treatment and a cerebral MRI was performed. We assessed the modified Rankin Score (mRS) at the time of each visit and angiography. Poor outcome was defined as a mRS≥3.
Results
Demographic data and clinical presentation
Between 2008 and 2013, 10 patients with angiographically visible dAVF of the anterior cranial fossa were consecutively treated by endovascular means in our institution. Patient baseline data and dAVF characteristics are described in table 1. Mean age was 59 years (range 46–73 years) with a male-to-female ratio of 0.67:1. The majority of dAVFs (n=4, 40%) were incidental findings and three (30%) were associated with another symptomatic cerebrovascular lesion. The clinical presentation was bleeding in two patients (20%) and seizures in one patient (10%). The initial WFNS score was 0 (no bleeding presentation) in seven cases (70%), II in one case (10%), III in one case (10%) and IV in one case (10%). The mRS at the time of the diagnosis was 0 in one patient (10%), 1 in four patients (40%), 2 in two patients (20%), 3 in one patient (10%) and 4 in one patient (10%).
Demographic data and dural arteriovenous fistulas characteristics
DAVF angio-architecture
All fistulas were duro-arachnoidal type between one or more ethmoidal artery and a frontal cortical vein. Ethmoidal branches from the ophthalmic artery were always involved, with a bilateral supply of the fistula in eight cases (80%). An additional supply by a branch of the middle meningeal artery was visible in three cases (30%) and by the internal maxillary artery in two cases (20%). The cortical vein involved in the drainage had a superficial course to the anterior third of the superior sagittal sinus (SSS) in six cases (60%) or a deep course to the anterior part of the cavernous sinus in two cases (20%). A superficial venous drainage by the superficial middle cerebral vein was noted in two cases. A venous ectasia was noted in seven cases (70%), which classified these fistulas as type IV. The three other cases were type III (30%). A venous reflux into other cortical veins was noted in only one case.
Embolisation through the ethmoidal artery
The first choice was always to embolise the fistula through the largest anterior ethmoidal branch via the ophthalmic artery. This technique was used or at least tried in 11 patients (10 embolisation sessions) with an efficiency rate of 45.5% (five cases). The embolic agent used was the Onyx in nine cases and the biological glue in two cases. The two embolisations with the biological glue were failures. In three cases (27.2%), a balloon (Scepter C (Microvention, Tustin, California, USA) in two cases and a Hyperglide (ev3 Neurovascular) in one case) inflated in the internal carotid artery was necessary to be able to catheterise the ophthalmic artery.
In two cases, the ophthalmic artery could not be catheterised despite the use of a balloon. The tip of the micro-catheter was placed in the third segment of the ophthalmic artery just above the origin of the anterior ethmoidal artery (figure 1). This was achieved without problem or complication with the different micro-catheters. The penetration of the embolic agent was sufficient in five cases. In four cases, arterial reflux of the embolic agent was too important and the procedure was discontinued. No entrapment of the micro-catheter tip happened during these treatments. No migration of the embolic agent was visible in another branch that originated on the first or second portion of the ophthalmic artery.
Lateral projection digital subtraction angiography (DSA) (A) showing the three segments of the ophthalmic artery (arrowheads) and the dilated anterior ethmoidal artery (*) in case of cribriform plate dural arteriovenous fistula. Working projection DSA (B) of balloon-assisted catheterisation of the ophthalmic artery ostium. Lateral projection DSA (C) with placement of the tip of the micro-catheter in the third segment of the ophthalmic artery. Selective injection DSA (D) of the fistula (lateral projection). Lateral view (E) during the injection of Onyx in the fistulous point and the draining vein. Control DSA in lateral projection (F) without visualisation of persistent early venous drainage.
Embolisation through the middle meningeal artery
In case of failure of the treatment by an ethmoidal artery and if the fistula was fed by a branch of the middle meningeal artery, we tried to use the anterior ramus of the middle meningeal artery. This technique was performed twice with a success rate of 50% (one case, figure 2). The middle meningeal artery has to be catheterised as close to the fistulous point as possible. The tortuosity of this artery made catheterisation impossible in one case. The penetration of the embolic agent was worsened by the long distance between the catheter tip and the fistulous point, and by the absence of artery flow due to rectification of the arterial course (figure 2). Despite the use of a balloon (Scepter, Microvention) for the cooker-pressure technique, the embolic agent did not penetrate the fistulous point. Arterial reflux was not important in all cases and no cranial nerve deficit was presented after the embolisation. No other post-treatment complication was noted.
Antero-posterior (A) and lateral (B) views of a selective injection of the middle meningeal artery presenting the tortuosity of the anterior branch (arrowheads). Working projections of selective injections of the middle meningeal artery with micro-catheter placement (D) and Onyx injection (D).
Embolisation by the venous approach
The venous approach was used in two cases. In one case it was after failure of the arterial approach, and in the second case, it was used to avoid navigation inside the ophthalmic artery in a patient already blind in the contralateral eye (figure 3). The two fistulas treated had a superficial venous drainage by the SSS with a short cortical distance. Success of the embolisation was noted in all cases without venous perforation. This venous approach was always done in a new embolisation session without the double catheter technique (arterial and venous). A distal catheter approach was always used to provide support and the microcatheter was a Sonic 1.5 (Balt, Montmorency, France) in one case and an Echelon 10 (ev3 Neurovascular) in the other. The embolic agent used was the Onyx in all cases, with the preliminary coiling of a venous ectasia in one case. The penetration of the fistulous point by the embolic agent was sufficient in all cases. No postoperative complication was noted and no venous ischemia was visible on follow-up MRI.
Illustrative case of venous approach. Lateral projection digital subtraction angiography (DSA) of right carotid artery showing a cribriform plate dural arteriovenous fistula with short and superficial cortical venous drainage. Catheterisation of the superior sagittal sinus (B) and selective injection (C) of the fistula in lateral projection. Successive injection of Onyx (D and E). Control DSA in lateral projection (F) without visualisation of persistent early venous drainage.
Failure of the endovascular treatment
In two patients (20%), endovascular therapy did not cure the fistula. The first case was a fistula only fed by one anterior ethmoidal artery with a long cortical venous distance. The only embolisation session consisted of arterial embolisation but, despite the use of a balloon in the internal carotid artery siphon, it was impossible to catheterise the ostium of the ophthalmic artery. The long cortical venous path with deep course was not favourable for a venous approach. The second case was a fistula supplied by an anterior ethmoidal artery and a middle meningeal branch, with a fronto-basal venous drainage coursing to the cavernous sinus. During the embolisation, we tried to treat the fistula by the middle meningeal artery but its tortuosity prevented distal navigation of the micro-catheter. Then we tried to catheterise the ophthalmic artery without success despite the placement of a balloon in the internal carotid artery. These two patients underwent a microsurgical exclusion of the fistula by the fronto-temporal approach. One presented a postoperative chronic subdural hematoma that was treated by two burr holes.
Clinical and ophthalmologic follow-up
The mean clinical follow-up was 13.1 months (range 4–60). Six patients presented a mRS 0, one a mRS 1 and three patients a mRS 2. The patient initially treated for seizures was seizure free at the last clinical follow-up without anticonvulsive medication. The two patients who presented with bleeding of the anterior fosse dAVF were mRS 2. One patient kept an abducens nerve palsy due to bleeding of a tentorial edge dAVF. All patients had an ophthalmologic examination during the postoperative course. The mean ophthalmologic follow-up was 5 months (range 2–9). No visual field loss was noted. Ocular tension was normal in all cases.
Angiographic follow-up
The mean angiographic follow-up was 9 months (range 4–13) after the last embolisation or surgery. No angiographic recurrence of the fistula was noted. Particular attention to thrombosis of ophthalmic artery branches has been made and no complication was seen.
Discussion
Cribriform plate dAVFs represent only a small proportion of cerebral arteriovenous fistulas (approximately 4–7% of all dAVFs in the literature and 3.7% in our own series). Based on a limited number of reports, the majority of cases seem to be treated by the microsurgical approach. Only a few isolated cases of endovascular-treated cases are reported in the literature, essentially for technical description of the treatment.
The microsurgical exclusion of the fistula is the first treatment and the most well known.13 ,14 The success rate of the surgical treatment is between 85% and 95%,4 ,14 with a rate of complication of 5.5–9% in experimented hands.3 ,14 Radiosurgical treatment is also described in isolated case reports in the literature.15 ,16 As for other cranial locations of dAVFs, radiosurgery has to be reserved for cases when open surgery and endovascular therapy are contraindicated because of low success rate and delayed cure.
Our series brings epidemiological and anatomical information to this pathology. Contrary to some authors, who appropriate a male predominance to the anterior fossa dAVFs,3 ,17 sex repartition is equal in our series with a discrete female predominance (60%). The age of discovery is similar to the majority of cases described in the literature, with a median age of 60 years,11 ,12 but the discovery mode is clearly different. Eighty percent of our cases are asymptomatic, contrary to other case reports which show a bleeding presentation in 50–74% of cases.3 ,6 ,7 ,12 In the literature,1 ,3–11 ,15 ,18 angio-architecture of cribriform plate dAVFs is identical to our findings (ethmoidal supply, often bilateral, cortical venous drainage).
Embolisation of cribriform plate dAVFs through an anterior ethmoidal artery is the technique of choice and the most described in the literature.1 ,3 ,19 ,20 The navigability of the last generation of micro-catheters allows us to catheterise the ophthalmic artery in most cases. The use of a balloon in the internal carotid artery siphon increased the success rate of catheterisation of the ophthalmic artery, as described by Zhao et al11 and Agid et al3 The ophthalmic artery is composed of three segments: the first segment extends from the optic canal to the point where the artery changes direction to cross over or under the optic nerve; the second segment is the short part of the vessel as it passes over or under the nerve; and the third segment extends from the bend in the vessel on the medial aspect of the optic nerve to the edge of the orbit. The central retinal artery originates from the second segment. The microcatheter is placed in the third segment of the ophthalmic artery just before the origin of the anterior ethmoidal artery. The penetration of the cribriform plate by the tip of the catheter is ideal but could not be achieved in all cases.9 ,13 ,19 The use of Onyx to embolise the fistula allows a good control of arterial reflux that must not be proximal to the junction between the second and third segments of the ophthalmic artery in order to protect the origin of the central retinal artery; and also a control of the penetration of the fistulous point and venous ectasia by the embolic agent. Short segment detachable catheters could be used to minimise traction on the ophthalmic artery upon removal of the catheter. The decreased inflow in the first and second segments of the artery after embolisation is always seen and is due to the absence of the fistula aspiration in an enlarged artery.
Branches of the middle meningeal artery could be used to treat the fistula. This technique is not well described in the literature.1 ,6 ,10 ,17 ,21 The success rate is low because of the tortuosity of the middle meningeal artery. As for other locations of dAVFs, this artery is difficult to navigate and the progression of the embolic agent is not as important as desired. The cooker-pressure technique could increase the penetration of the embolic agent without predictable effect.22
The venous approach to treat dAVFs presents a higher complication rate than the arterial approach.3 ,8 ,17 ,18 In anterior fossa dAVFs, the venous path has to be well evaluated. A long cortical distance and deep path to the cavernous sinus or lateral path to the superficial middle cerebral vein are contraindications for the use of the venous approach. In the case of a short and superficial venous pathway, the venous approach shows high effectiveness. If the access to the SSS is endovascularly impossible, a mini-craniotomy in front of the SSS has to be considered to access it.
With a combination of all these endovascular techniques, the rate of success could be acceptable without exposing the patient to unnecessary risks. Endovascular therapy could be considered as first-line treatment for this type of dAVF. In case of failure of endovascular techniques, open surgery has to be considered because of the high risk of bleeding of duro-arachnoidal fistulas.
Conclusion
In our series, endovascular treatment of cribriform plate dAVF seems to be safe and effective with a success rate of 80% without ophthalmic and neurological complications. We should note that the low number of patients limits the results of our series. Embolisation through the ethmoidal artery is the method of choice. Branches of the middle meningeal artery are often too tortuous and prevent the penetration of the embolic agent. The venous approach is effective but has to be considered only after failure of the arterial approach and in cases with superficial and short cortical venous drainage.
References
Footnotes
Contributors TR: writing and design of the study. GC, RF and SS: data analysis. HR and BB: lecturer. RB and MP: design of the study, supervision.
Competing interests None declared.
Patient consent Obtained.
Ethics approval Rothschild Foundation Hospital Ethical Committee.
Provenance and peer review Not commissioned; externally peer reviewed.