Cavernous sinus dural arteriovenous malformations

Cavernous sinus dural arteriovenous malformations

Cavernous Sinus Dural Arteriovenous Malformations Patterns of Venous Drainage are Related to Clinical Signs and Symptoms Hadas Stiebel-Kalish, MD,1 Av...

172KB Sizes 9 Downloads 45 Views

Cavernous Sinus Dural Arteriovenous Malformations Patterns of Venous Drainage are Related to Clinical Signs and Symptoms Hadas Stiebel-Kalish, MD,1 Avi Setton, MD,2 Yassunari Nimii, MD,2 Yuval Kalish, MA,3 Jonathan Hartman, MD,2 Ruth Huna Bar-On, MD,4 Alejandro Berenstein, MD,2 Mark J. Kupersmith, MD5 Objective: To provide evidence that venous congestion and drainage patterns are responsible for the manifestations of cavernous sinus area dural arteriovenous malformations (CSdAVMs). Design: Retrospective observational case series. Participants: Records of 85 patients with complete clinical and angiographic evaluations of CSdAVMs were evaluated for the clinical features of the disorder. A neuroradiologist analyzed patterns of venous drainage to and from the cavernous sinus without knowledge of the clinical features. Four venous drainage patterns (reversal of flow from the CSdAVMs into the anterior cavernous sinus, ophthalmic vein thrombosis, drainage into the inferior petrosal sinus or drainage into the superior petrosal sinus) were statistically tested for their predictive value of signs and symptoms using logistic regression. Main Outcome Measures: The power of prediction of orbital congestion, elevated IOP, extraocular muscle dysfunction, optic neuropathy, venous-stasis retinopathy, choroidal effusion, anterior chamber shallowing, bruits, cranial nerve paresis, and central nervous system dysfunction from four patterns of venous drainage. Results: Reversal of drainage into the anterior cavernous sinus and ophthalmic veins was highly predictive (P ⫽ 0) of orbital congestion, which was seen in 77 (91%) patients. In contrast, eight (9%) patients without orbital congestion had shunts that did not drain into the anterior cavernous sinus and ophthalmic veins. Cavernous sinus dural arteriovenous malformation drainage into the anterior cavernous sinus and ophthalmic veins also predicted elevated IOP (P ⫽ 0.0023) and optic neuropathy (P ⫽ 0.047). Ophthalmic vein thrombosis significantly predicted cases with choroidal effusion (P ⫽ 0.002) and anterior chamber shallowing (P ⫽ 0.01). Third nerve paresis could be predicted from flow toward the inferior petrosal sinuses (P ⫽ 0.017). Central nervous system symptoms or dysfunction, occurring in 7 (8%) patients, was predicted by venous drainage into the superior petrosal sinus (P ⫽ 0.0008). Conclusions: The clinical features found in patients with CSdVAMs are related to the abnormal venous drainage and can be predicted by these venous drainage patterns. Venous congestion and hypertension seem to cause the clinical dysfunction in this disorder. Ophthalmology 2002;109:1685–1691 © 2002 by the American Academy of Ophthalmology. Cavernous sinus dural arteriovenous malformations (CSdAVMs) cause a well-known spectrum of signs and symptoms including proptosis and orbital congestion, diplopia

Originally received: November 1, 2002. Accepted: April 5, 2002. Manuscript no. 200743. 1 Neuro-Ophthalmology Service, Department of Ophthalmology, Rabin Medical Center, Petah Tikva, Israel. 2 Endovascular Service, Beth Israel Medical Center, New York, New York. 3 Department of Psychology, University of Melbourne, Melbourne, Australia. 4 Neuro-Ophthalmology Service, Sheba Medical Center, Ramat Gan, Israel. 5 Neuro-Ophthalmology Service at Institute of Neurology and Neurosurgery, Beth Israel Medical Center, New York Eye and Ear Infirmary, and NYU School of Medicine, New York, New York. Correspondence to Hadas Stiebel-Kalish, MD, Neuro-Ophthalmology Service, Department of Ophthalmology, Rabin Medical Center, Petah Tikva 49100, Israel. © 2002 by the American Academy of Ophthalmology Published by Elsevier Science Inc.

both from congestive ocular motor dysfunction and from cranial nerve paresis, increased intraocular pressure (IOP), and retinal hemorrhages and edema resembling venous stasis retinopathy.1 Visual loss can also develop because of optic neuropathy, secondary glaucoma, and choroidal effusion or detachment? Few patients suffer a more aggressive neurologic deterioration caused by intracereberal hemorrhages and cerebral venous infarct. Because CSdAVMs are not typically high flow, and the involved arteries continue to adequately supply the orbit and cranial nerves, the extent and type of signs and symptoms are not readily associated with specific arterial supply of these shunts. Ophthalmic symptoms in CSdAVMs are infrequently caused by arterial steal.2–5 Normal and abnormal patterns of cavernous sinus venous outflow are detailed in Table 1 and depicted in Figure 1. The anterior part of the cavernous sinus drains blood from the common ophthalmic vein. The sphenoparietal sinus and the ISSN 0161-6420/02/$–see front matter PII S0161-6420(02)01166-1


Ophthalmology Volume 109, Number 9, September 2002 Table 1. Normal and Abnormal Cavernous Sinus Outflow Direction Normal

Abnormal (“reversal”)

Venous Channel Posterior Inferior Opposite cavernous sinus Anterior Superior Posterior

deep and superficial middle cerebral veins also empty into the anterior cavernous sinus. The posterior cavernous sinus receives blood from the anterior cavernous sinus.1 From the anterior cavernous sinus, blood flows both into the posterior cavernous sinus and inferiorly to the pterygoid venous plexus and through emissary veins of the foramen rotundum and foramen ovale, located in the floor of the middle cranial fossa. The blood in the posterior cavernous sinus normally

Superior and inferior petrosal sinuses Emissary veins to foramen rotundum, foramen ovale, foramen of Vesalius Anterior and posterior coronary sinuses, basal venous plexus Ophthalmic venous system Sphenoparietal sinus, middle cerebral vein Via superior petrosal sinus to ponto-mesencephalic veins

drains into the inferior and superior petrosal sinuses. The inferior petrosal sinus drains into the jugular vein at the latter’s junction with the sigmoid sinus junction. The superior petrosal vein empties where the transverse and sigmoid sinuses meet. Blood can also flow into the opposite cavernous sinus via the anterior and posterior coronary sinuses and through the basal venous plexus, which lies on the clivus, posterior to the dorsum sella.1

Figure 1. Diagram of venous drainage to and from the cavernous sinus. ac ⫽ anterior coronary sinus (connecting the cavernous sinuses); bvr ⫽ basal vein of Rosenthal; cs ⫽ cavernous sinus (larger than true proportion); fo ⫽ vein to foramen ovale (to pterygoid plexus); fr ⫽ vein to foramen rotundum (to pterygoid plexus); fv ⫽ vein to foramen venosum (Vesalius’ vein); ic ⫽ internal cerebral vein; ips ⫽ inferior petrosal sinus; ocs ⫽ opposite cavernous sinus; ov ⫽ ophthalmic vein; pc ⫽ posterior coronary sinus (connecting the cavernous sinuses); pmv ⫽ ponto-mesencephalic veins; smc ⫽ superficial middle cerebral vein; spars ⫽ sphenoparietal sinus; sps ⫽ superior petrosal sinus; ss ⫽ straight sinus; sss ⫽ superior sagittal sinus; sv ⫽ sylvian vein; vg ⫽ vein’s of Galen.


Stiebel-Kalish et al 䡠 Neuro-ophthalmologic Findings and Venous Drainage in CSdAVMs This study was designed to provide quantitative evidence that specific patterns of venous drainage can predict the clinical orbital and neurologic findings in patients with CSdAVMs and predict that these abnormal venous drainage patterns are responsible for the majority of clinical manifestations of these patients.

Materials and Methods The records of 139 patients with CSdAVMs between January 1981 and July 1998 were reviewed and evaluated by MJK at the neuroophthalmology services at the Institute of Neurology and Neurosurgery, Beth Israel Medical Center, the New York Eye and Ear Infirmary, and the New York University School of Medicine. Selected from this group were 85 patients with dural arteriovenous malformations (dAVMs) of the cavernous sinus region for whom complete clinical and angiographic evaluations were available. Excluded from the study group were 13 cases of diffuse dAVMs of the tentorium or transverse sinus that drained to various structures and only indirectly to the cavernous sinus, 38 patients who did not undergo angiographic examination (most because endovascular therapy was not warranted), and three patients whose CSdAVMs had spontaneously closed before angiography (so that a correlation between venous drainage and eye signs could not be examined). Patients with carotid cavernous fistulas, a single-hole arteriovenous (AV) shunt supplied from a branch of the cavernous internal carotid artery (ICA), were not included in this study. All 85 patients had AV shunts supplied principally, but not exclusively, by dural arteries. Bilateral selective cerebral angiography of the external carotid artery (ECA) and ICA was used to determine if the CSdAVM was supplied from the ECA or ICA branches or from both systems. Superselective catheterization was necessary to demonstrate all of the arteries supplying each CSdAVM and the potential collaterals between the ECA and ICA systems, as well as the safest, most expedient route for treatment. Vertebral angiography was also performed whenever necessary to complete the evaluation. The directions of the shunt blood flow into and away from the cavernous sinus were readily demonstrated. Normal and abnormal cavernous sinus outflow patterns are detailed in Table 1 and Figure 1. Notation was made of stagnant flow (delayed emptying of contrast dye) or thrombosis (lack of flow) in the ophthalmic veins. A drainage route was considered to be dominant when preferential flow or marked venous dilation was observed, whereas less dilation or normal caliber suggested less flow. Abnormal (reversal of) flow was noted when the cavernous flow was opacified after injection into the ECA.2 All of the venous phase of the angiogram also was used to demonstrate whether there was abnormal flow toward the cortical veins determined by abnormally dilated cortical veins or dye filling. A neuroradiologist who had no knowledge of the clinical findings determined the characteristics of venous outflow. The neuro-ophthalmic records were analyzed for details of clinical signs and symptoms. These findings were consistently and prospectively documented by MJK for each patient; clinical findings were recorded in a uniform fashion to be analyzed in the future. Patients were considered to have orbital congestion if they had arterialized conjunctival vessels and at least one of the following additional signs: lid edema, conjunctival chemosis or proptosis, and extraocular muscle local dysfunction. Some of these patients with orbital congestion also had, in various combinations, increased IOP, anterior chamber narrowing, choroidal effusion, and venous-stasis retinopathy (prominent retinal venous dilatation with at least one of the following: retinal edema, retinal exudates,

retinal hemorrhages, disc hyperemia, or central vein occlusion). Optic neuropathy was diagnosed when a relative afferent pupillary defect and color vision loss (by pseudoisochromatic plate testing), usually with reduction in visual acuity and a detectable visual field defect, were present. Extraocular muscle underaction caused by local congestion was distinguished from cranial nerve dysfunction by the former having other signs of orbital congestion and normal speed of saccades by clinical observation, although the ocular excursions were restricted. We assumed that clinically apparent slowed saccades were more likely to reveal a neuropathic process caused by cranial neuropathy, and that normal saccades would be more likely to be present with extraocular muscle congestion, despite the fact that these were not absolute but clinical estimate characteristics. These results were correlated with the patterns of abnormal venous drainage. P values for each separate predictor were considered to be significant only if the entire model was found to be significant (P ⬍ 0.05).

Results Of the 85 patients, there were 34 men and 51 women. Average age at presentation was 65.5 years in women and 60.7 years in men. Thirteen patients had true bilateral AV shunts, and 72 patients had unilateral AV shunts. No patient had filling of the AV shunt via the ophthalmic artery before the appearance of dye in the distal orbital arteries and choroidal blush, and therefore, there were no cases of arterial steal into the AV shunt. Selective ICA angiography showed normal filling of the supracavernous circulation, even when branches of the cavernous ICA supplied the CSdAVM.

Venous Drainage Patterns The patterns of venous drainage seen on cerebral angiography of the 85 patients in this study are summarized in Table 2. All 13 patients with bilateral CSdAVMs had venous drainage into both cavernous sinuses and into both right and left ophthalmic venous systems. Of the 72 patients with unilateral shunts, 12 patients had venous drainage into both cavernous sinuses and ophthalmic venous systems, 58 patients had a unilateral dAVM draining into its ipsilateral cavernous sinus and ophthalmic vein, and two patients had a unilateral dAVM draining into the contralateral cavernous sinus through the interconnecting coronary sinus. Bilateral arterial supply, from both right and left dural branches of the ICA or ECA on both sides, was seen in 33 of the 72 patients with unilateral CSdAVMs. Seventy-seven patients had CSdAVMs that drained into the anterior cavernous sinus and ophthalmic venous system. Absence of flow in the ophthalmic vein due to preferential flow away from the ophthalmic venous system was seen in eight cases. Eleven patients had angiographic evidence of ophthalmic vein thrombosis. Seven other patients had stagnant flow in the ophthalmic vein ipsilateral to the shunts’ drainage. Venous drainage into the inferior petrosal sinus occurred in 22 patients; in six of these patients this flow was dominant (see Materials and Methods for definition of dominant flow). Eleven CSdAVMs drained into the superior petrosal sinus, two of these having their dominant flow in this direction. Twenty-two of the 85 (26%) patients had cortical venous drainage.


Ophthalmology Volume 109, Number 9, September 2002 Table 2. Four Main Patterns of Venous Drainage on Cerebral Angiography of 85 Patients with Cavernous Sinus Dural Arteriovenous Malformations* Venous Outflow from Cavernous Sinus Pattern of Venous Drainage

No. of Patients

Additional Venous Drainage Patterns

1. Reversal of flow into the anterior cavernous sinus and ophthalmic vein


2. Abnormal ophthalmic venous flow

a. Ophthalmic vein thrombosis


b. Stagnant (slow) ophthalmic vein flow 3. Inferior petrosal sinus



4. Superior petrosal sinus


Cortical veins–19 (25%) Inferior petrosal sinus– 16 (21%) Superior petrosal sinus–9 (12%) Cortical veins–8 (73%) Inferior petrosal sinus– 3 (27%) Superior petrosal sinus–3 (27%) Cortical veins–1 (14%) Superior petrosal sinus–0 Inferior petrosal sinus–0 Cortical veins–9 (41%) Superior petrosal sinus– 4 (18%) Paraspinal veins–1 (4.5%) Cortical veins–11 (100%)

*See “Materials and Methods.”

The Incidence of Clinical Signs in 85 Patients with CSdAVMs Table 3 lists the incidence of clinical signs in this series of 85 patients with CSdAVMs.

Signs of orbital congestion were seen in 77 (91%) patients. A unilateral red eye and congested orbit were seen in 52 (61%) patients and bilateral congestion occurred in 25 (29%). Eight patients had no signs of orbital congestion (“white-eyed” shunt syndrome). Ocular motor dysfunction, resulting from local extraocular muscle congestion, was observed in 20 of the 85 patients in this series. All 20 patients with motility limitation also had signs of orbital congestion (by definition). Seven patients were observed to have coexisting cranial nerve dysfunction and additional restriction due to extraocular muscle congestion (20 patients had extraocular muscle restriction, 35 had cranial nerve paresis, but only 7 had both, differentiated as explained in Materials and Methods). Increased IOP was seen in 61 (72%) patients, despite maximally tolerated medical treatment. The majority of these patients were treated with ocular hypotensive agents, including topical ␤-blockers, topical ␣-adrenergic agents, and topical or systemic carbonic anhydrase inhibitors. While under treatment, mild elevation of IOP (20 –25 mmHg) was observed in 20%, moderate elevation (25–30 mmHg) in 24%, and a rise above 30 mmHg on maximally tolerated medication in 28% of eyes. Ocular hypertension did not occur without concomitant signs of orbital congestion. Glaucomatous cupping and corresponding visual field defects were observed in 11 of the 61 patients with raised IOP. Anterior segment ischemia occurred in a small number of patients (nine) and did not permit valid statistical analysis. These patients had, in varying combinations, anterior chamber cells or flare (four patients), corneal edema (three patients), iris neovascularization (one patient), and iris sphincter ischemia (three patients). Hypoxic-ischemic venous-stasis retinopathy occurred in 44 (52%) patients. These patients had, in varying combinations, prominent retinal venous dilatation (30 patients), scattered retinal hemorrhages (15 patients), and optic disc edema (7 patients). Ten

Table 3. Frequency of Clinical Signs in 85 Patients with Cavernous Sinus Dural Arteriovenous Malformations Clinical Signs

Total No.

Side of signs


Orbital congestion


Secondary glaucoma


Anterior chamber shallowing Choroidal effusion Hypoxic-ischemic retinopathy

11 7 44

Cranial neuropathy


Subjective bruit Objective bruit

24 7

Neurological signs and symptoms



No. Unilateral Bilateral No ocular signs Proptosis Lid edema Arterialized conjunctival vessels Conjunctival chemosis Extraocular muscle local dysfunction Increased intraocular pressure Glaucomatous cupping and field loss Venous dilation Retinal hemorrhages Optic nerve edema Retinal edema Retinal exudates Central vein occlusion Optic neuropathy Sixth nerve paresis Third nerve paresis Fourth nerve paresis Fifth nerve dysfunction Facial nerve paresis Orbital Auricular Vertigo Intracerebral hemorrhage Cortical venous infarct

57 25 3 65 49 65 18 20 61 11 11 7 30 15 7 10 1 1 26 29 16 4 2 1 3 4 4 2 1

% of 85 67 29 4 76 58 76 21 24 72 13 13 8 35 18 8 12 1 1 31 34 19 5 2 1 28 4 5 5 2 1

Stiebel-Kalish et al 䡠 Neuro-ophthalmologic Findings and Venous Drainage in CSdAVMs patients had retinal edema (nine who had decreased visual acuity and one who also had retinal exudates). One patient had a clinical picture of a central vein occlusion. Eleven patients had anterior chamber shallowing; all of these patients also had orbital congestion. This occurred in the context of clinically detectable choroidal effusion in 3 of the 11 patients. However, patients with anterior chamber shallowing were not routinely dilated, so the incidence of choroidal effusion could have been higher. Six of the 11 patients with anterior chamber shallowing also had increased IOP. Choroidal effusion was clinically detected in seven patients (three who had elevated IOPs ⬎30 mmHg and four who had normal IOPs). All patients with choroidal effusion also had orbital congestion. Twenty-six patients had an optic neuropathy. These patients all had orbital congestion. Dysfunction of the third, fourth, first division of the fifth, sixth, or seventh cranial nerves occurred in 35 patients. Paresis of the sixth cranial nerve was the most common, which was observed in 29 (34%) patients. The third cranial nerve was involved in 16 (19%) patients. Fourth nerve paresis was seen in three patients and was never seen without other ocular motor nerve deficits. Presence of fourth nerve paresis in the presence of third nerve dysfunction was appreciated by the presence of intorsion when the patient was requested to look down and in. Concomitant paresis of the third and sixth nerves was seen in seven (8%) patients. Facial nerve paresis occurred in one patient who also had signs of orbital congestion and symptoms of dizziness. Seven patients had features of both restrictive extraocular limitation and cranial nerve dysfunction. Of the eight patients with “white-eyed” shunts, five presented with cranial nerve dysfunction, two had a sixth nerve deficit, two had combined sixth and third nerve involvement, and one had a complete pupillary involving third nerve deficit. The other three patients with “white-eyed” CSdAVMs presented with neurologic symptoms and signs (the latter two patients presented with intracerebral hemorrhage, the former with persistent dizziness), but without any eye signs. The two patients with sixth nerve paresis did not have microvascular risk factors nor any other readily related cause for their dysfunction. After magnetic resonance imaging scans performed as part of their diagnostic work-up, these two patients were referred for angiography which revealed radiographic evidence of a CSdAVM. The three patients with “whiteeyed” shunts also had abnormal magnetic resonance imaging scans as part of their evaluations before referral, which led to the diagnosis of the CSdAVMs. Twenty-four (28%) patients experienced a subjective bruit that could not be detected by an examiner. An objective bruit that could be heard by the examiner was found in seven (8%) patients. The bruit was heard over an orbit in three patients and behind the ear in four. Central nervous system symptoms or dysfunction occurred in seven patients. Of the seven patients, four had symptoms of vertigo and dizziness, whereas three suffered central nervous system (CNS) dysfunction as a result of a venous infarct or from an intracerebral hemorrhage. These 7 patients, and the 15 asymptomatic patients had cortical venous drainage. Intracerebral hemorrhage occurred in two patients who had ectatic and dilatated cortical veins. A venous infarct occurred in one patient with CSdAVM flow into dilatated cerebellar veins. True vertigo occurred in one patient with perimesencephalic drainage. Three other patients complained of dizziness, one who had perimesencephalic drainage and two who had drainage into the sylvian vein, presumably disturbing the temporal lobe. Seven of the 22 patients with cortical venous drainage had CNS symptoms or deficits, whereas none of the patients without cortical drainage had CNS symptoms. The affected cortical veins appeared tortuously dilated and had

delayed emptying on angiography. After closure of the shunt, arteriography demonstrated restoration of the normal cerebral circulation. On repeat angiography several weeks to months after treatment, the previously arterialized veins filled in normal sequence, and drained in the proper direction, and their caliber was normalized. The CSdAVM of the three patients with no orbital or cranial nerve signs drained mainly into the inferior or superior petrosal sinuses. One of these patients underwent angiography after an intracerebral hemorrhage caused by cortical venous drainage from his CSdAVM (mentioned previously). The second patient only had a persistent bruit. The third patient presented with severe dizziness and was found to have a high-grade carotid stenosis on Doppler; and cerebral arteriography revealed the asymptomatic CSdAVM.

Analysis of Correlation between Venous Drainage Patterns and Clinical Signs in CSdAVMs Of the 25 patients with bilateral eye signs, 10 patients had true bilateral arterial venous shunts with each dAVM draining into the ipsilateral cavernous and ophthalmic venous systems, whereas 11 others had a unilateral shunt draining into both cavernous sinuses and resulting in congestion of both orbits. Four patients had bilateral eye signs without angiographic evidence of venous drainage into both ophthalmic veins. Three patients had unilateral orbital congestion although they had bilateral shunts that drained into both ophthalmic veins. One patient with unilateral cranial neuropathy and no orbital congestion had a unilateral AV shunt with bilateral ophthalmic venous drainage.

Clinical Signs Predicted by CSdAVM Drainage into the Anterior Cavernous Sinus and Ophthalmic Vein Reversal of flow from the CSdAVM into the anterior cavernous sinus and the ophthalmic vein perfectly predicted all patients with orbital congestion (P ⫽ 0; odds ratio [OR], infinitesimal). These results make logistic regression analysis of orbital congestion, as predicted from the other venous drainage patterns, impossible because of overfitting. Other signs predicted by anterior cavernous sinus and ophthalmic venous drainage included elevated IOP (P ⫽ 0.002; OR, 29.81) and presence of optic neuropathy (P ⫽ 0.047; OR, 3.96). Anterior cavernous sinus and ophthalmic vein drainage was found to be a near significant negative predictor of CNS signs or symptoms (P ⫽ 0.055; OR, 0.118).

Clinical Signs Predicted by Ophthalmic Vein Thrombosis Ophthalmic vein thrombosis could significantly predict cases with anterior chamber shallowing (P ⫽ 0.01; OR, 9.2) and choroidal effusion (P ⫽ 0.002; OR, 25.33).

Clinical Signs Predicted by CSdAVM Drainage into the Inferior Petrosal Sinus Shunt drainage into the inferior petrosal sinus could predict cases with third cranial nerve paresis (P ⫽ 0.017; OR, 2.59).

Clinical Signs Predicted by Drainage of CSdAVMs into the Superior Petrosal Sinus Drainage into the superior petrosal sinus significantly predicted the seven patients with CNS symptoms or signs (P ⫽ 0.001; OR, 27.6).


Ophthalmology Volume 109, Number 9, September 2002

Discussion The predominant pattern of abnormal venous drainage in CSdAVMs, reversal of ophthalmic venous flow, and resultant ophthalmic venous hypertension can predict the presence of orbital congestion, secondary glaucoma, and optic neuropathy (see Results for specific P values). When there are no clinical signs of ophthalmic venous hypertension, such as with the “white-eyed” shunt syndrome, the abnormal venous drainage pattern is predominant in the inferior or superior petrosal sinuses, whereas the ophthalmic venous system is completely or relatively spared. Because we have found no case of increased IOP or optic neuropathy in a patient with a “whiteeyed” shunt, we believe that unilateral glaucoma in a noncongested orbit, or an unexplained optic neuropathy without orbital congestion, is unlikely to be caused by a CSdAVM. Bilateral eye signs usually occurred with venous drainage into both anterior cavernous sinuses and ophthalmic venous systems, whereas unilateral eye signs typically occurred in cases with ipsilateral anterior cavernous drainage. We cannot fully explain how four patients had bilateral orbital findings with only unilateral ophthalmic vein drainage. Possibly, there was prior bilateral anterior cavernous sinus drainage, or perhaps we were unable to see thrombosis in the anterior cavernous sinus or ophthalmic vein that prevented the appearance of abnormal venous drainage. Ophthalmic vein thrombosis was found to predict two ocular complications of CSdAVMs (i.e., anterior chamber narrowing and choroidal effusion). This prediction is especially striking given the relatively small number of cases with these complications. Thus, the physician should probably not consider using the transophthalmic venous approach to close the AV shunt in these cases.6 Cavernous sinus dAVMs that caused third nerve dysfunction were more likely to drain, in addition to the ophthalmic venous system, toward the inferior petrosal sinus. In contrast, the AV shunts that did not cause cranial nerve paresis usually drained preferentially into the ophthalmic venous system. Other authors have ascribed cranial nerve paresis to arterial steal, with blood shunted away from nutrient arteries to the affected cranial nerves or to a thrombus-caused mass effect compression of the nerve against the dural wall of the cavernous sinus.7– 8 However, we did not find a high incidence of cavernous sinus thrombosis in cases with cranial neuropathy. Because five patients (age range, 36 – 83 years) had a cranial nerve dysfunction without orbital congestion, a CSdAVM should be considered in the differential diagnosis of a persistent neuropathy. Facial nerve paresis occurred in one patient who also had signs of orbital congestion and symptoms of dizziness. This patient had cortical venous drainage and possible perturbation of the facial nerve from an adjacent congested venous sinus or steal toward the shunt from the nerve’s dural blood supply in the petrous region (middle meningeal artery branches).1 Subjective bruits were experienced by 28% of patients in this series, whereas 8% had objective bruits that could be detected by stethoscope in the orbital or postauricular area. Objective postauricular area bruit occurred in only 5% of patients and was associated with shunt drainage into the inferior petrosal sinus; but the model for this predictor was not


significant, possibly because of the very small number of cases with this bruit. Central nervous system symptoms and dysfunction were observed in 8% of the patients in this series. Central nervous system signs and symptoms could be predicted by CSdAVM drainage into the superior petrosal sinus, which then drains into cortical veins. All of these cases had tortuous and dilated cortical veins draining blood flow away from the CSdAVM. Altogether, 26% of our patients had cortical venous drainage. Because approximately one third of this subgroup were symptomatic, this finding on angiography or magnetic resonance imaging seems to be an important risk factor in the development of cerebral venous hypertension, venous infarct, and intracereberal hemorrhage. This observation is in concert with the findings of Cognard et al9 and Awad et al10 who reported that dAVMs with increased risk for CNS dysfunction are those with cortical venous drainage and ectasia. It is important to remember that the venous occlusive process of CSdAVMs is a dynamic one, so patterns of drainage can change along with deterioration of neuro-ophthalmic course or disappearance of signs and symptoms.11–12 For example, one of our patients with sixth nerve paresis as the only sign of CSdAVM experienced a spontaneous resolution of signs and of the shunt. Another patient with a sixth nerve paresis initially improved, then developed a venous infract with signs of expressive aphasia. A third patient presented with a devastating intracerebral hemorrhage several months after spontaneous resolution of a sixth nerve paresis. In reading the angiogram, reliance on a single neuroradiologist may produce errors. We attempted to minimize such interobserver errors by asking the reading neuroradiologist (AS) to interpret the angiogram without knowledge of the existing written report. This report had been interpreted previously by other members of the neuroradiology service (AB and YN). Any differences in interpretation were brought for a third reading. Because CSdAVMs tend to be slow or low flow shunts, we were not surprised to find no angiographic evidence of a steal from the ophthalmic artery or the ICA above the shunt. Several authors have previously described cases of cavernous area dAVMs presenting with amaurosis fugax caused by arterial steal from the ophthalmic artery.2–5 New blood flow measurement techniques may reveal slow or reduced arterial flow into the eye, or it is possible that these patients with retinal or choroidal dysfunction actually had slow outflow or transit time into and out of the ophthalmic venous system, as in ischemic oculopathy. In our series, the retinal and choroidal signs and symptoms seemed to be related to the abnormal pattern of venous drainage and resultant venous hypertension in the orbit and eye, as well as thrombosis in the ophthalmic veins.

References 1. Kupersmith MJ. Neurovascular Neuro-Ophthalmology. New York: Springer-Verlag, 1993;64,125–35. 2. Vinuela F, Fox AJ, Pelz DM, Drake CG. Unusual clinical manifestation of dural arteriovenous malformation. J Neurosurg 1986;64:554 – 8.

Stiebel-Kalish et al 䡠 Neuro-ophthalmologic Findings and Venous Drainage in CSdAVMs 3. Bogousslavsky J, Vinuela F, Barnett HJM, Drake CG. Amaurosis fugax as the presenting manifestation of dural arteriovenous malformation. Stroke 1985;16:891–3. 4. Xiong L, Li J, Jinkins JR. Amaurosis fugax caused by dural arteriovenous fistula from the ophthalmic artery. AJNR Am J Neuroradiol 1993;14:191–2. 5. Fritz W, Klein HJ, Schmidt K. Arteriovenous malformation of the posterior ethmoidal artery as an unusual cause of amaurosis fugax. The ophthalmic steal syndrome. J Clin Neuroophthalmol 1989;9:165– 8. 6. Kupersmith MJ, Vargas EM, Warren F, Berenstein A. Venous obstruction as the cause of retinal/choroidal dysfunction associated with arteriovenous shunts in the cavernous sinus. J Neuroophthalmol 1996;16:1– 6. 7. Lasjaunais P, Chiu M, ter Brugge K, et al. Neurological manifestations of intracranial dural arteriovenous malformations. J Neurosurg 1986;64:724 –30.

8. Hawke SHB, Mullie MA, Hoyt WF, et al. Painful oculomotor nerve palsy due to dural-cavernous sinus shunt. Arch Neurol 1989;46:1252–5. 9. Cognard C, Gobin YP, Pierot L, et al. Cerebral dural arteriovenous fistulas: clinical and angiographic correlation with a revised classification of venous drainage. Radiology 1995; 194:671– 80. 10. Awad IA, Little JR, Akarawi WP, Ahl J. Intracranial dural arteriovenous malformations: factors predisposing to an aggressive neurological course. J Neurosurg 1990;72:839 – 50. 11. Sergott RC, Grossman RI, Savino PJ, et al. The syndrome of paradoxical worsening of dural-cavernous sinus arteriovenous malformations. Ophthalmology 1987;94:205–12. 12. Kupersmith MJ, Berenstein A, Choi IS, et al. Management of nontraumatic vascular shunts involving the cavernous sinus. Ophthalmology 1988;95:121–30.