Treatment of 54 traumatic carotid-cavernous fistulas

Transcription

1 J Neurosurg 55: , 1981 Treatment of 54 traumatic carotid-cavernous fistulas GI~RARD DEBRUN, M.D., PIERRE LACOUR, M.D., FERNANDO VINUELA, M.D., ALLAN FOX, M.D., CHARLES G. DRAKE, M.D., AND JEAN P. CARON, M.D. Department of Neurosciences, University Hospital, London, Ontario, Canada, and Universitaire Henri Mondor, Paris, Cr~teil, France ~/ A series of 54 traumatic carotid-cavernous fistulas has been treated with detachable balloon catheters. The balloon was introduced through one of three different approaches: the endarterial route; the venous route through the jugular vein, the inferior petrosal sinus, and the cavernous sinus; or surgical exposure of the cavernous sinus; with occlusion of the fistula by a detachable balloon directly positioned in the cavernous sinus. Full follow-up review demonstrated that the carotid blood flow was preserved in 59% of cases. The most frequent complication was a transient oculomotor nerve palsy, which occurred in 20% of cases. In three cases where both the fistula and the carotid artery were originally occluded by the balloon, the superior portion of the fistula was later found not to be completely occluded, and these patients had intracranial ligation of the supraclinoid portion of the carotid artery. Three patients had hemiparesis, transient in two cases and permanent in the other. The results show that the fistula was totally occluded in 53 cases; in the one exception the patient became asymptomatic but had a minimal angiographic leak. KEY WORDS 9 carotid-cavernous fistula 9 embolization 9 therapeutic embolization 9 balloon catheter occlusion T RAUMATIC carotid-cavernous fistulas can be treated in different ways. 1,3,5-s,15-31 Permanent occlusion of the fistulas and the carotid artery with a Fogarty catheter 6,7,1~ is gradually being replaced by more sophisticated techniques, all of which aim to occlude the fistula while preserving carotid blood flow. 1,3,5,9,13,15-18,20,22,23,25,31-33 These new techniques use different approaches. In the first, a detachable balloon catheter is introduced via an endarterial route, and enters the cavernous sinus through the hole in the fistula. The balloon is inflated and detached in the cavernous sinus to occlude the fistula, leaving the carotid artery patent. This technique has been described by Serbinenko 31,32 and Debrun, et al., 9 and subsequently by others. 3,5,~3,23,25,33 They used latex detachable balloons, but Silastic detachable balloons are now available. 1~ The endarterial route was also used by Kerber t7 and Bank, et al., 1 who have successfully sealed some carotid-cavernous fistulas with a calibrated-leak balloon and bucrylate. The second alternative is to introduce the balloon catheter via a venous route to the cavernous sinus. This is an excellent way to treat a carotid-cavernous fistula that drains anteriorly through the superior ophthalmic vein 22,26 or posteriorly through the inferior petrosal sinus, 2~ and it avoids the risks of intraarterial manipulations with a coaxial system. A few 678 cases have been treated through this route with Fogarty catheters, but it is more satisfactory to detach a balloon into the cavernous sinus and to preserve the inferior petrosal sinus and the jugular vein. The third alternative is the surgical approach to the cavernous sinus. Parkinson 24 has surgically occluded the rent of a fistula with the patient under cardiac arrest and deep hypothermia. Hosobuchi, et al., t6,35 reported the use of electrothrombosis of the cavernous sinus. Mullan 22 described various ways of inducing thrombosis of the cavernous sinus with thrombogenic needles or bronze wire. In this approach, the cavernous sinus is punctured anteriorly through the superior ophthalmic vein or posteriorly in Parkinson's triangle or through the superior petrosal sinus. A direct injection of bucrylate into the cavernous sinus has been used by a few people, but the risk of reflux of the bucrylate into the carotid artery is high, and control of the fistula occlusion is difficult at surgery. We describe a new method. Patient Population Summary of Cases Seventeen cases were treated at the Hfpital Universitaire Henri Mondor from 1974 to The J. Neurosurg. / Volume 55/November, 1981

2 Treatment of traumatic carotid-cavernous fistulas results in these patients have been published previously. 9 Thirty-seven other cases were treated at University Hospital, London, Ontario, Canada, from August, 1978, to February, The age and sex distribution in this series of 54 patients is shown in Fig. 1. There was a strong prevalence of young male patients in the second and third decades. Most of the older patients were female. In total, there are 33 males and 21 females. Of the 54 patients, 53 had known head injury, caused in almost all cases by a motor-vehicle accident. Three older patients had a fall, and the youngest patient, a 5-year-old girl, was hit in the eye with a fishing rod. One patient, a sailor, did not remember any injury, but his fistula had all the characteristics of a traumatic etiology, having a big rent, a high flow, and no participation of the external carotid artery. He has therefore been included in this series. Angiographic Studies The precise location of the fistula was identified in all 54 cases. In 51 cases, vertebral angiography with compression of the carotid artery in the neck on the side of the fistula showed retrograde filling of the internal carotid artery (ICA) through the posterior communicating artery. The carotid artery below the fistula did not fill, and it was easy to locate and measure the size of the rent (Fig. 2). In two cases in which the posterior communicating artery was too small, the exact location was identified by means of a double-lumen balloon catheter. 2,4 The balloon was inflated in the ICA just above the bifurcation, and iodine contrast material was slowly injected through the other lumen. The contrast medium filled the ICA above the balloon and showed the origin of the fistula FIG. 1. Bar graph showing age and sex of the 54 patients in this series. (Fig. 3). In one case treated 15 years ago by trapping of the carotid artery and division of the carotid bifurcation in the neck, a hypoglossal artery reconstituted the ICA and filled the fistula. Once the flow through the fistula was reduced, one film of the routine ICA seriography showed the exact location and size of the rent. A summary of the incidence of fistulas at each location in the intracavernous ICA is given in Fig. 4. Following the ICA from the anterior clinoid process to the petrous canal, the fistula was located: 1) on the anterior ascending intracavernous segment of the carotid artery in one case; 2) at the junction of the anterior ascending and horizontal segment of the carotid artery in five cases; 3) on the horizontal intracavernous portion of the carotid artery in 22 cases; 4) at the junction of the horizontal and intracavernous ascending segment of the carotid artery in 15 cases; and 5) on the posterior ascending intracavernous segment in 11 cases. FIG. 2. Left: Angiogram showing a right traumatic carotid-cavernous fistula. Right: Demonstration of the location of the fistula by vertebral arteriography with compression of the carotid artery in the neck. Retrograde filling of the internal carotid artery is seen. The rent is precisely demonstrated. J. Neurosurg. / Volume 55 / November,

3 G. Debrun, et al. The fistulas varied in size from 1 to 5 mm or more, with an average of 3 mm. The fact that one balloon closed the fistula in almost all cases proved that there was only one rent. However, in two cases we had proof that there was more than one rent. In the first case, the fistula was initially occluded with a detachable balloon, but 2 weeks later the fistula recurred and angiography showed a second hole lower than the first; the first fistula was still occluded by the first balloon. In the second case, one balloon entered the cavernous sinus through one hole and a second balloon entered through another hole. There was one case of bilateral fistulas in this series, an extremely rare occurrence that has been reported in very few cases, s,n FIG. 3. Angiogram demonstrating the location of the fistula and the position of the double-lumen balloon catheter. The balloon (open arrow) occludes the internal carotid artery. Iodine contrast material is injected slowly through the balloon and demonstrates the rent in the fistula (black arrow). Total Steal Total steal (that is, complete absence of filling of the ICA above the fistula) was present in seven cases. Total steal did not mean that the ICA was occluded above the fistula: vertebral angiography and contralateral carotid angiography with ipsilateral compression of the carotid artery in the neck filled the fistula in a retrograde direction, proving that the ICA was still patent (Fig. 5). Circle of Willis The communicating arteries were generally widely patent in these cases. Hypoplastic anterior and pos- FIG. 4. Number of cases with fistulas occurring at various locations of the intracavernous internal carotid artery. 680 J. Neurosurg. / Volume 55 / November, 1981

4 Treatment of traumatic carotid-cavernous fistulas terior communicating arteries were seen in one case, and in another case, treated many years previously with ligation of the ICA in the neck, vertebral angiography showed no posterior communicating artery. In the latter, the anterior communicating artery was patent and filled the ICA in a retrograde direction and then the fistula, but the first segment, M1, of the middle cerebral artery did not t111 on the side of the fistula. In fact, this patient presented with transient ischemic attacks that disappeared after occlusion of the fistula. In all the other cases, there was filling of either the posterior communicating, or the anterior communicating artery, or both. Role of External Carotid Artery The external carotid arteries did not supply the fistula in 53 of the 54 traumatic cases of this series. The one exception was a traumatic case that was treated 10 years before by division of the external and internal carotid artery in the neck. The ICA had not been ligated intracranially, and the fistula still fed from the contralateral carotid through the anterior communicating artery. The external carotid artery filled from the vertebral artery through the occipital artery, and gave off branches to the ophthalmic artery and to the fistula. Venous Drainage The venous drainage can be anterior, posterior, inferior, superior, or contralateral. Anterior, posterior, and contralateral venous drainage is often associated. Anterior venous drainage through the superior ophthalmic vein is frequent, inducing proptosis and chemosis. The more anterior the fistula, the more prominent the anterior venous drainage. However, in cases of partial thrombosis or incomplete filling of the cavernous sinus, anterior venous drainage can be totally absent, even with an anterior fistula (Fig. 3); or venous drainage can be exclusively anterior (Fig. 6 left). Fro. 5. Upper Left: Angiography in a patient showing traumatic carotid-cavernous fistula with total steal. Upper Right: Contralateral carotid arteriography showing the internal carotid artery filling in a retrograde direction above the fistula. Lower: Vertebral arteriography with compression of the carotid artery. There is retrograde filling of the carotid artery and of the fistula. Posterior venous drainage is often associated with anterior venous drainage. The inferior petrosal sinus can be the only posterior venous drainage (Fig. 3), but there may also be drainage through the superior petrosal sinus. The more posterior the fistula, the more prominent the posterior venous drainage. Proptosis and chemosis associated with these fistulas can be minimal. Contralateral venous drainage through the coronal veins is frequently associated with ipsilateral venous drainage. When contralateral venous drainage is predominant, proptosis and chemosis can be contralateral as well. It is exceptional to find that the only venous drainage is via the contralateral cavernous sinus (Fig. 6 right). Superior venous drainage through the superficial Sylvian vein and collaterals and inferior venous drainage through the pterygoid plexus are sometimes associated with the other types of drainage. The superficial Sylvian venous network can be well developed in cases with fistulas and a total steal. J. Neurosurg. / Volume 55 / November,

5 G. Debrun, et al. TABLE 1 Treatment and results in 54 cases of traumatic carotid-cavernous fistulas* Case No. Age (yrs), Sex Approach Oculomotor Neurological Venous Arterialt Surgical Palsy Complications Fistula Carotid False Occluded Preserved Aneurysm 1 25, M ND 2 ND , M ND 1 ND , F ND 3 ND - transient hemiparesis 4 23, M ND 2 ND , M ND 2 ICA exposure - - in neck 6 33, M ND l ND , M ND 1 ND , M ND 1 ND , M ND 1 ND - - l0 29, M ND l ND , M ND 1 ND , M ND 2 ND , F failure l ND , M ND 1 ND , F ND l ND , M ND 1 ND , M ND 2 ND , F ND l ND , M ND 2 ND , F ND 3 ND , M ND l ND , F failure 2 ND transient 7th nerve palsy 12 days after treatment 23 22, M ND 3 ND , M ND 1 ND , M failure 2 ND , M failure 2 ND , M ND 2 ND , F ND 1 ND , F ND l ND , F ND l ND , F ND 2 ND , F ND l ND , F ND 1 ND , M ND 1 ND , M ND 1 ND , M successful ND ND , M ND 1 ND , M ND 2 ND , F ND balloon de- failure of plug- - - tached into ging cav sinus cav sinus with muscle 40 68, F ND l ND , F ND 3 intracran permanent ligation hemiparesis 42 19, F ND 1 intracran - - ligation 43 25, F ND 4 ND , M ND 1 ND , F ND 1 ND , M 2 balloons: no ND ICA ligation; - - occlusion Bucrylat injection 47 29, M 1 balloon: no 1 ND - occlusion 48 5, F ND 1 ND - - small small small - big~ - big~ - big$ - big~ small previous - surgical ligation - big~: - small previous - ligation * ICA = internal carotid artery; ND = not done; - = absent; = present; cav = cavernous. t Number of operations. :~ These patients had a big false aneurysm which had to be treated by permanent occlusion of the ICA because of intractable retro-orbital pain or oculomotor nerve palsy. These five patients had at least two operations. 682 J. Neurosurg. / Volume 55 / November, 1981

6 Treatment of traumatic carotid-cavernous fistulas TABLE 1 (continued) Case Age Approach Oculomotor Neurological Fistula Carotid False No. (yrs), Sex Venous Arterialt Surgical Palsy Complications Occluded Preserved Aneurysm , M 82, F ND 1 balloon: no 1 failure to en- ND 2 balloons de- - died 3 wks - occlusion ter cav tached into later: sepsis sinus cav sinus 51 16, M failure 1 ligation of intra cranial ICA 52 51, M 1 balloon: no failure to en- plugging with occlusion ter cav muscle sinus 53 18, M failure failure Bucrylat into cav - transient - - sinus hemiparesis 54 50, F ND 1 ND * ICA = internal carotid artery; ND = not done; - = absent; = present; cav = cavernous. t Number of operations. These patients had a big false aneurysm which had to be treated by permanent occlusion of the ICA because of intractable retro-orbital pain or oculomotor nerve palsy. These five patients had at least two operations. Clinical Presentation It is not our purpose to describe the clinical symptomatology of these patients in great detail. These patients were examined by an ophthalmologist before treatment. The vision in the involved eye and the oculomotor nerve palsies, if any, were carefully noted and compared to the posttherapeutic status. The degree of proptosis and chemosis correlated well with the venous drainage through the superior ophthalmic vein. On the other hand, patients with minimal proptosis and chemosis but with a loud bruit were likely to have dominant posterior venous drainage. The interventional neuroradiologist must auscultate the patient and note the degree of murmur in the neck, in the temporal region, and over the eye. Disappearance of the murmur over the eye is a good clinical indication that the balloon is occluding the fistula. The results of compressing the carotid artery in the neck on the side of the fistula must be carefully interpreted. If the patient tolerates the compression, he will probably tolerate well a permanent occlusion of the carotid artery if this becomes necessary. A more objective manner to obtain this information is to occlude the ICA below the fistula with the balloon catheter for 5 minutes at the time of treatment. However, if the patient does not tolerate the compression and becomes hemiplegic and unconscious after a few seconds, this response proves that he cannot tolerate both occlusion of the carotid artery and the ongoing steal through the fistula. Such a patient will probably tolerate compression of the carotid artery after treatment of his fistula with a balloon and preservation of the carotid blood flow. Mode of Treatment The approach to treatment of the fistula can be either endarterial, venous, or surgical. Table 1 sum- marizes the mode of treatment and results in this series of patients. Endarterial Balloon Catheterization Treatment. Fifty-two patients were treated by endarterial balloon catheterization, 45 exclusively and seven in association with another approach, either venous or surgical. The 17 patients managed in France were treated through the neck, and the 37 patients operated on in Canada were treated through the groin in 32 cases and through the neck in five cases. Each route has advantages and disadvantages. The approach through the neck requires a balloon catheter half the length of the catheter needed in approaching through the groin, and the dead space of the short catheter is less than half the dead space of the long one. When the balloon is inflated with silicone, the quantity of iodine that is reintroduced into the balloon when silicone is injected is smaller with a short catheter than with a long one. Some fistulas are occluded with a balloon inflated with 0.02 or 0.03 ml, which is much less than the capacity of the dead space of 0.10 ml of the shortest catheter. With a single-lumen catheter, silicone can be used in more cases, and we have more control of the coaxial system with the neck approach because the catheter is shorter. Conversely, there are many disadvantages. It is difficult to safely introduce a No. 8 French introducer into the carotid artery. Dissection of the intima can result, and the carotid artery can become spastic or thrombose. There is a risk of cervical hematoma when the introducer is removed, and it is safer to carry out the whole procedure while the patient is intubated. The dose of radiation received by the neuroradiologist is also increased with this method, because we work very close to the source of x-rays. The approach through the groin is generally easier J. Neurosurg. / Volume 55 / November,

7 G. Debrun, et al. FIG. 6. Angiograms showing a traumatic carotid cavernous fistula with exclusively anterior venous drainage (left). The venous drainage is also exclusively contralateral (right). and safer, but some patients with tortuous arteries cannot be easily catheterized with a straight No. 8 French introducer, even if it is exchanged with a regular catheter using a 250-cm guide. The main inconvenience is the greater number of patients in whom silicone cannot be used because of the increased dead space of a single-lumen catheter. This happens whenever the useful capacity of the balloon needed to occlude the fistula is smaller or equal to the dead space of the catheter (0.2 ml). Balloons that are inflated only with iodine contrast material (iso-osmolar to blood or not) and detached, shrink progressively and are totally deflated after 1 month. If they deflate too quickly, a venous pouch or false aneurysm will occur. Results. All fistulas treated via the endarterial approach were occluded except one, in which there is still a minimal leak although the patient is asymptomatic. The carotid blood flow was preserved in 32 (59%) of the total 54 cases. The reasons for permanent occlusion of the ICA are summarized in Table 2. A venous pouch or false aneurysm occurred in 24 (44%) of the 54 cases. When this pouch is small, it remains asymptomatic for years and does not alter in size (Fig. 7). Big pouches, between 1 and 2 cm in diameter, generally occur a few days after occlusion of the fistula because the balloon has deflated too quickly (Fig. 8 right). The fistula continues to be occluded, but the false aneurysm develops in place of the inflated balloon, and the pouch is the same size as the initial size of the inflated balloon. Big pouches occurred when the balloons were inflated with 1.0 ml or more of iodine. If the balloon has been partially filled with silicone, the pouch will be smaller. The pouch can induce intractable retro-orbital pain or oculomotor nerve palsy. If this complication occurs, it is necessary to permanently occlude the carotid artery and the neck of the false aneurysm (that is, the rent of the original fistula) with a second detachable TABLE 2 Reasons for permanent internal carotid artery (1CA) occlumon in 22 patients in this series Reasons for ICA Occlusion No. of Case Nos. Cases failure to enter cavernous sinus 5 (2) big false aneurysm complicated by oculomotor nerve palsy or intractable retro-orbital pain 5 or both cavernous sinus could be entered but fistula could not be occluded 4 previous surgical ICA ligation in neck and/or intracranially 2 balloon occluded carotid artery below fistula but a leak persisted that had to be treated by 3 intracranial ligation of ICA & ophthalmic artery ICA thrombosis due to dissection of artery with the guide 1 ICA thrombosis during procedure 1 ICA stenosis by bulging balloon associated with a traumatic ICA dissection in neck in a 1 motor-vehicle accident; carotid artery thrombosed on follow-up angiography 2, 13, 28, 33, 40 (Cases 4l & 42 included below) 19, 20, 22, 26, 43 24, 34, 38, 48 39, 46 41,42, 51 (in Cases 41 & 42 it was also impossible to enter cavernous sinus) J. Neurosurg. / Volume 55 / November, 1981

8 Treatment of traumatic carotid-cavernous fistulas FIG. 7. Angiography in a patient with traumatic carotid-cavernous fistula treated with an iodineinflated balloon. Left: One month after treatment, a false aneurysm has developed in place of the balloon, which has progressively deflated. A small incidental ophthalmic aneurysm can be seen. Right: One year later, the false aneurysm is smaller. The piece of silver inside the deflated balloon is still visible (arrow). balloon. The pain disappears instantaneously and the oculomotor nerve palsy resolves in a few weeks. We had to occlude the carotid artery as a second step in five cases. In three other cases, it was possible to de- No. of Carotid False Carotid Oculomo- Artery. Aneu- Artery tor Nerve tach a second balloon in the pouch and to cure the Cases Preserved rysms Stenosis Palsy patient. Table 3 compares the results achieved with silicone- balloons inflated with silicone and iodine-inflated balloons, indicating that better 14. lo balloons inflated with iodine results are obtained with the former (Fig. 9). The four false aneurysms that occurred in the series of 14 patients with silicone-filled oalloons were due to a technical complication that prevented us from deliv- TABLE 3 Comparison of results with silicone- and iodine-inflated balloons * One patient had been treated previously with an iodine-filled balloon, then developed a false aneurysm that was occluded with a second balloon inflated with silicone. FIG. 8. Left: Angiography showing a traumatic carotid-cavernous fistula treated with two balloons (two silver clips and arrow) inflated with iodinated material. The fistula is occluded and the carotid artery is preserved. Right: Three weeks later, the balloons have deflated and there is a big false aneurysm. This patient had intractable retro-orbital pain at this time, and the carotid artery was permanently occluded with a third balloon. The patient recovered completely. J. Neurosurg. / Volume 55 / November,

9 G. Debrun, et al. FIG. 9. Angiogram showing a right traumatic carotid-cavernous fistula. Left: Balloon inflated with a polymerizing substance (silicone). Right: Excellent result 6 months later. ering the precise amount of silicone into the balloon. Correct use of silicone-filled balloons should dramatically decrease the number of false aneurysms, but might increase the incidence of stenosis of the carotid artery. Oculomotor nerve palsy was the second most prevalent serious complication, occurring in 11 (20%) of the 54 cases (Table 4). It was frequently found when more than one balloon was detached in the cavernous sinus, occurring in 50% of these cases. The palsy occurred within hours of treatment and was due to excessive compression of the nerve by the inflated balloons. As a rule, the first balloons were all inflated with iodine contrast material, and the last balloon, which occluded the rent of the fistula, was inflated with silicone. Iodine-filled balloons progressively deflated, and the oculomotor nerve palsy recovered more easily when the pressure inside the cavernous sinus decreased. Sixth nerve palsy occurred more often and recovered faster than third nerve palsy. Patients recovered from their oculomotor nerve palsy reasonably well after some weeks or months. There was only one case of permanent partial nerve palsy. The fistula recurred in five cases. In one case, the balloon moved inside the cavernous sinus and a second balloon had to be detached 3 days later; however, the clinical result was good and the carotid blood flow was preserved. In another case (Case 53), a second rent of the carotid artery opened below the first one, necessitating a second surgical treatment, because the second balloon could not enter the cavernous sinus. In the three other cases (Cases 41, 42, and 51), it was impossible to enter either the cavernous sinus or the carotid artery, as the fistula was occluded with a detachable balloon. In these cases, the balloon had deflated slightly after detachment and the fistula had recurred, filling in a retrograde direction from above through the communicating arteries or the ophthalmic artery. These three patients had intracranial clipping of the ICA and of the ophthalmic artery. One of these patients (Case 41), a 76-year-old woman, had a middle cerebral artery infarction with a permanent neurological deficit. The two other patients recovered totally. The only other neurological complication seen in this group was a transient facial nerve palsy that occurred 12 days after treatment in Case 22. Venous Approach The venous approach can be performed surgically or by catheterization. Whenever the cavernous sinus drains through a large inferior petrosal vein, it is advisable to try to reach the cavernous sinus through the jugular vein and inferior petrosal sinus. The jugular vein is easily catheterized with an introducer from the femoral vein or by direct puncture in the neck. The inferior petrosal sinus is catheterized with the balloon catheter. Our detachable balloon technique allows us to reach the cavernous sinus and detach the balloon using the femoral vein approach. Another catheter is positioned in the carotid artery to confirm if the fistula is occluded or not. In fact, the number of patients who can be successfully treated through this venous approach is very small, because the partitions of the cavernous sinus often prevent the balloon from reaching the chamber close to the rent of the fistula. The balloon catheter can be moved against the flow, but 686 J. Neurosurg. / Volume 55 / November, 1981

10 Treatment of traumatic carotid-cavernous fistulas FIG. 10. Left: Angiogram of a left traumatic carotid-cavernous fistula with posterior venous drainage. Center: Vertebral arteriography with compression of the carotid artery. The small posterior communicating artery (small arrow) faintly opacities the rent in the fistula (big arrow) and the cavernous sinus. Right: One balloon was introduced through the inferior petrosal sinus (direct puncture of the left jugular vein) and detached into the cavernous sinus. Four months after treatment, there is a minimal residual false aneurysm (arrow). the narrow openings between several chambers of the cavernous sinus prevent easy passage. Among 12 patients undergoing the venous approach (Table 5), the cavernous sinus was reached with the balloon in five cases, and the balloon totally occluded the fistula in Case 36 (Fig. 10). One or several balloons were detached in three cases (Cases 46, 47, and 52), but did not completely occlude the fistula, which was treated by endarterial catheterization in one case and by surgical procedure in the two other cases. In the fifth patient, an 82-year-old woman (Case 50), the balloon occluded the venous drainage through the petrosal sinus, which increased the anterior venous drainage and the proptosis. Catheterization via the endarterial route was attempted, but the balloons did not enter the cavernous sinus. The patient was unlikely to tolerate occlusion of the carotid artery. The patient developed a third nerve palsy in the hours following the procedure, and her vision deteriorated casc. TABLE 4 Summary of oculomotor nerve palsies in 11 cases Involved No. of No. of Balloons Nerve Cases Detached* 3rd alone 3 4, 8, 1 (with a big pouch) 6th alone 6 1, 2, 4, 2, 2, 3 3rd & 6th 2 1 (with a big pouch, both cases) * Number of balloons detached into the cavernous sinus in each TABLE 5 Results of venous approach in 12 cases* Venous Case Arterial Surgical Approach No. Approach Treatment one balloon detached into cavernous sinus occluded fistula one or several balloons detached into cavernous sinus but did not occlude fistula 47 successful cause of previous surgical ligation failure to enter inferior petrosal sinus with balloon catheter 46 impossible be- intracranial ICA ligation & ophthalmic artery inj of bucrylate below 52 unsuccessful cavernous sinus plugged with muscle 50 unsuccessful 2 balloons detached into cavernous sinus 13 successful 22 successful 25 successful 26 successful 51 incomplete occlu- ligation of intrasion of fistula by cranial ICA, of balloon detached ophthalmic ar- into cavernous 1CA 54 successful tery, & inj of bucrylate into siphon below failure to catheterize superior ophthalmic vein 53 unsuccessful injection of bucrylate into cavernous sinus * ICA = internal carotid artery; inj = injection. J. Neurosurg. / Volume 55 / November,

11 G. Debrun, et al. dramatically. She was operated on the day after, and the cavernous sinus was punctured with a No. 7 French introducer. Two balloons were introduced into the cavernous sinus and detached, and the fistula was occluded. The carotid artery was partially stenosed but patent on follow-up angiography 5 days later. Unfortunately, this patient died 3 weeks later of septic complications. The superior ophthalmic vein is sometimes enormous and drains into a wide angular facial vein. However, all the attempts to reach the cavernous sinus in a retrograde fashion through the facial and angular vein failed because of the sharp angles of the vein at the level of the trochlea of the superior oblique muscle (Case 53). Surgical Approach Two patients in this series were treated by detachable balloons introduced into the cavernous sinus at surgery. The first patient was the same 82-year-old woman who was previously treated with a detachable balloon introduced through the inferior petrosal sinus (Case 50). The second of these patients was a 45-yearold woman (Case 39) who had had a right traumatic carotid-cavernous fistula treated 15 years ago by division of the carotid artery in the neck and intracranial clipping of the ICA. She had enucleation of her right eye. In spite of the previous treatment, her fistula was not occluded; she continued to hear a loud bruit and developed a tremendous dilatation of the veins of the forehead and upper eyelid. Angiography showed reconstitution of the ICA from the vertebral artery through a hypoglossal artery (Fig. 11). The fistula was posterior, and the only venous drainage was via a dilated superior ophthalmic vein. The posterior communicating artery was still patent, and contrast medium opacified the ICA and then the fistula. The occipital artery was widely anastomosed with the vertebral artery and reconstituted the external carotid artery. Two modes of treatment were discussed. The first was catheterization of the vertebral artery or occipital artery, hypoglossal artery and ICA with a calibratedleak balloon. If the balloon had reached the cavernous sinus, injection of bucrylate (Bucrylat*) could have sealed the fistula. The second possibility was surgical exposure of the cavernous sinus and plugging it with pieces of muscle. Both of these different types of procedures failed, the first because the calibrated-leak balloon never entered the hypoglossal artery and the fistula still filled after the second procedure. A venous approach to the cavernous sinus through the superior ophthalmic vein was thought to be hazardous because the fistula was located posteriorly. We were also reluctant to inject bucrylate into the cavernous sinus * Bucrylat manufactured by Ethicon Co., 1421 Lansdowne Street West, Peterborough, Ontario, Canada. because of the risk of reflux into the vertebral artery. A further alternative was a new surgical approach to the cavernous sinus. The posterior communicating artery was clipped, and then the cavernous sinus was punctured with a No. 7 French sheath. A balloon, inflated with 1.0 ml of Conray 60, was detached into the cavernous sinus. The murmur over the eye became inaudible, and on follow-up angiography the fistula had disappeared. The patient has been free of symptoms for 8 months. Occlusion of the fistula was entirely successful in the two cases where it was attempted. Radiological Evaluation Discussion During the radiological evaluation of the fistula, we happened to note that there was a complete steal phenomenon with no tilling at all of the ICA beyond the fistula in seven cases. In all these cases, vertebral angiography or contralateral carotid angiography with ipsilateral compression of the carotid artery always demonstrated that the ICA above the fistula was patent and tilled the fistula in a retrograde direction. Some think that in the presence of total steal, permanent occlusion of the ICA without trying to enter the cavernous sinus with the balloon is the treatment of choice. If we agree that total steal indicates that the patient tolerates the absence of flow above the fistula well, we think it is a mistake to treat this type of fistula differently. However, we must recognize that the treatment will generally be more difficult in these cases, in which a big cavernous sinus often requires occlusion with more than one balloon. Many authors have suggested that bilateral external carotid angiography should be carried out in all cases of traumatic carotid cavernous fistula, but we think it is unnecessary in most cases. The absence of contribution by the external carotid branches is so constant that we think it is the best indication of the traumatic origin of the fistula. A comparative series of 16 spontaneous carotid-cavernous fistulas studied during the same period of time showed, on the contrary, a contribution by the external carotid artery in all except one case of ruptured cavernous aneurysm. Global Results The overall results of this series show that the fistula was occluded in all cases but one, in which a minimal leak was still seen on follow-up angiography but no clinical symptoms were evident. The carotid artery was preserved in 59% of the 54 cases, but the artery was seldom totally normal. There was often a minor stenosis or a small false aneurysm which was asymptomatic in all the cases, This abnormality often diminishes in size, as we have noted on a certain number of follow-up angiograms at 1 year (Fig. 5). 688 J. Neurosurg. / Volume 55/November, 1981

12 Treatment of traumatic carotid-cavernous fistulas FIG. 11. Angiography of a traumatic carotid-cavernous sinus fistula treated 15 years before by division of the carotid artery in the neck and intracranial ligation of the internal carotid artery (ICA). Upper Left: The vertebral artery reconstitutes the external carotid, which reconstitutes the ICA. A hypoglossal artery anastomoses the vertebral artery to the ICA. Upper Right: Later phase of the vertebral angiography series. The superior ophthalmic vein is the only means of venous drainage of the fistula. Lower Left: A balloon was detached at surgery into the cavernous sinus and occluded the fistula. Lower Right: Vertebral angiography 1 week after treatment showing no Idling of the ICA. Complications Symptomatic false aneurysms are the big pouches that develop very soon after treatment due to rapid deflation of an iodine-filled balloon as a result of an imperfect ligature. The number of these big pouches should be reduced in the future by the use of silicone and a double-lumen catheter. In the five cases where this complication occurred, we had to treat the patient a second time to permanently occlude the carotid artery. It is amazing to note how fast after occlusion of the carotid artery pain disappears and oculomotor nerve palsy resolves. J. Neurosurg. / Volume 55 / November,

13 G. Debrun, et al. The incidence of false aneurysms was high (44%) and was partially due to technical factors. The large number of patients treated through the groin in this series (32 cases) increased the number of cases where silicone could not be used to inflate the balloon, because of the dead space of the catheter. The capacity of the balloon needed to occlude the fistula is often equal to or smaller than the capacity of the dead space of the single-lumen catheter and whenever we injected silicone, we reintroduced into the balloon the amount of iodine trapped in the dead space of the catheter. The use of a double-lumen catheter should solve this problem, but this catheter is bigger and stiffer than the single-lumen variety, and it is more difficult to maneuver within the artery and to introduce the balloon into the cavernous sinus. Each lumen of this double-lumen catheter is also extremely small, and it takes a long time to inflate and deflate the balloon, which increases the duration of the whole procedure. For the present, the extensive use of silicone requires either that a double-lumen catheter is utilized, despite its drawbacks, or that the length of the singlelumen catheter is reduced as far as possible by inserting it through the neck. We must anticipate an increase in the incidence of stenosis of the carotid artery when the balloon bulges into the siphon through the fistula, and perhaps an increase in the number of oculomotor nerve palsies which will not completely recover. This is due to the difference between a silicone-filled balloon, which will maintain its size and shape permanently, and an iodine-filled balloon, which will progressively shrink after a few weeks, reducing the pressure inside the cavernous sinus but causing a false aneurysm. Oculomotor nerve palsy developed in 20% of the cases and represents a severe complication of the treatment. This problem is due to the compression of the nerve by the balloon inside the cavernous sinus. It is advisable to avoid inflating more than one balloon in the cavernous sinus with silicone. The iodineinflated latex balloons can shrink whereas siliconeinflated balloons keep their size permanently. It is curious that this complication has not been mentioned by those who have extensively used detachable balloons in the cavernous sinus. Fortunately, these oculomotor palsies recover after some weeks or months. Other neurological complications included a transient peripheral seventh nerve palsy that occurred 15 days after treatment of the fistula. Probably the inflammatory reaction in and around the cavernous sinus compromised the blood supply of the facial nerve, unless it was a coincidental Bell's palsy. The patient recovered completely. Another patient developed a middle cerebral artery infarction with permanent neurological deficit. This complication occurred after surgical intracranial ligation of the carotid artery, necessitated by incomplete occlusion of the fistula by the balloon detached in the carotid artery. Two other patients suffered transient hemiparesis; in one this occurred at surgery during injection of bucrylate into the cavernous sinus, when some bucrylate escaped through the fistula and embolized the middle cerebral artery. This is the only case which was treated that way, but emphasizes the risk involved in treating a carotid-cavernous fistula with bucrylate either at surgery by direct injection into the cavernous sinus or through the arterial route with a calibratedleak balloon. Kerber 17 reported three cases of carotidcavernous fistula treated with his calibrated-leak balloon and the injection of bucrylate. He noted that all three patients suffered significant but temporary neurological deficit either during or after the treatment, but later became asymptomatic. Approaches for Balloon Catheterization The three different approaches, endarterial, venous, and surgical, are today three alternatives that can be used alone or together. There are two different periods in this series. From 1974 to 1978, the venous approach was tried only once and failed. From 1978 to the time of writing, the venous approach has always been tried first whenever the fistula was posterior and the inferior petrosal sinus filled. The venous route is extremely effective and safe. Mullan 22 was probably the first to treat a carotidcavernous fistula this way, with occlusion of the fistula and preservation of the carotid blood flow. The few cases published in the literature, including Mullan's case, have been treated with a Fogarty catheter which is permanently buried in the soft tissues of the neck. As the balloon cannot be detached, this technique has to be performed through the jugular vein in the neck. Our detachable balloon technique allows us to use the femoral vein, and to detach the balloon into the cavernous sinus. The only patient who was successfully treated this way had a normal ICA at angiography 4 months after the procedure.t It must be emphasized that this route is extremely elegant, but that the percentage of success is very low because of the partitions of the cavernous sinus and the difficulty of reaching the rent of the fistula with the balloon. The other endovenous catheterization procedure that theoretically could be attempted is retrograde catheterization of the superior ophthalmic vein when the fistula is anterior. This vein anastomoses widely with the angular vein and the facial vein. To our knowledge, Peterson, et al., z~ are the only surgeons who have reported reaching the cavernous sinus with a thrombogenic bronze wire and treating the fistula successfully through this approach. Our balloon catheter failed to pass the sharp angles of the superior ophthalmic vein in one patient (Case 53), who was t Since writing this paper, we have successfully treated a second case through the inferior petrosal sinus. 690 J. Neurosurg. / Volume 55 / November, 1981

14 Treatment of traumatic carotid-cavernous fistulas finally treated by injection of bucrylate into the cavernous sinus at surgery. It is generally emphasized that the venous approach avoids all the complications of the endarterial route. However, it must be noted that to control occlusion of the fistula, an arterial catheter should be placed into the carotid artery during the procedure with the accompanying risk of emboli. To avoid this danger, we have systematically heparinized our patients, whatever route, venous or arterial, was used. When the venous and arterial approaches fail to occlude the fistula and preserve the carotid artery, is it wiser to permanently occlude the carotid artery with a detachable balloon or to propose a surgical approach to the cavernous sinus? Permanent occlusion of the ICA with a detachable balloon that occludes both the fistula and the artery is an easy, fast, and usually safe procedure, similar to treatment with Prolo or Fogarty catheter. There is no risk of oculomotor nerve palsy, because the balloon does not penetrate the cavernous sinus. The only risk is if the balloon occludes the artery below the fistula but does not completely occlude the fistula. This complication occurred in three of our cases. It is a serious complication because the hemodynamics of the circle of Willis have been modified in such a way that the patient can develop ischemic symptoms on the side of the fistula, because the carotid artery is occluded but the steal phenomenon through the fistula is still present. Immediate intracranial ligation of the internal carotid and ophthalmic artery is the best treatment to avoid any ischemic event and to cure the patient who would otherwise continue to have all the symptoms of his fistula. This risk and the fact that it is illogical to cure a fistula by permanent occlusion of an artery has led us to favor the surgical approach. The surgical approach was initially proposed by Parkinson, 24 but was not used by many neurosurgeons because occlusion of the rent of the fistula under cardiac arrest and deep hypothermia was considered a major and risky procedure for a patient who could be treated more easily and safely in other ways. Currently, the tremendous improvements in surgical technique and microdissection open a new field which will have a great future. Mullan's 22 different techniques of introducing thrombogenic needles or wires into the cavernous sinus, anteriorly or posteriorly, are effective and elegant. Hosobuchi '~ has reported electrothrombosis of the cavernous sinus. In the present series, Drake has cured several carotid-cavernous fistulas by plugging the cavernous sinus with pieces of muscle; the puncture of the cavernous sinus with a No. 7 French introducer that he used in two cases of this series has allowed us to detach one balloon in one case and two balloons in the other, with complete occlusion of the fistula and preservation of the carotid blood flow. Our experience with the surgical approach is too recent and too limited to claim that all those patients whose carotid artery we occluded would have fared better with surgery. Our hope is that the combination of venous catheterization, arterial catheterization, and surgical exposure of the cavernous sinus will considerably reduce the number of cases in which the carotid artery has to be permanently occluded. References 1. Bank WO, Kerber CW, Drayer BP, et al: Carotid cavernous fistula: endarterial cyanoacrylate occlusion with preservation of carotid flow. J Neuroradiol 5: , Berenstein A, Kricheff II: Balloon catheters for investigating carotid cavernous fistulas. Radiology 132: , Brunon J, Duquesnel J, Fischer G: [Free percutaneous embolization, with preservation of carotid flow of a traumatic carotid-cavernous fistula. Apropos of a case in a patient with a traumatic aneurysm of the intracranial carotid treated by "trapping."] Neuroehirurgie 23: , 1977 (Fre) 4. Cares HL, Roberson GH, Grand W, et al: A safe technique for the precise localization of carotid-cavernous fistula during balloon obliteration. Technical note. J Neurosurg 49: , Chermet M, Cabanis EA, Debrun G, et al: [Carotidcavernous fistula treated with inflatable balloons.] Bull Soe Ophtalmol Fr 77: , 1978 (Fre) 6. Chowdhary UM: Treatment of carotid-cavernous fistula using a balloon-tipped intra-arterial catheter. 3 Neuroi Neurosurg Psychiatry 41: , Cophignon J, Djindjian R, Rey A: ]Treatment of carotid-cavernous fistulae.] Ann Radiol 17: , 1974 (Fre) 8. Dardenne G: Treatment of bilateral carotid-cavernous fistulas. J Neurosurg 44:268, 1976 (Letter) 9. Debrun G, Lacour P, Caron JP, et al: Detachable balloon and calibrated-leak balloon techniques in the treatment of cerebral vascular lesions. J Neurosurg 49: , Devlamynck S, Rossazza C, Jan M, et al: [Ocular complications sustained after the occlusion of a carotidcavernous fistula with a Fogarty balloon catheter.] Rev Otoneuroophtalmol 50: , 1978 (Fre) 11. Donnell MS, Larson S J, Correa-Paz F, et al: Traumatic bilateral carotid-cavernous sinus fistulas with progressive unilateral enlargement. Surg Neurol 10: , Duffer JL, Merland J J, Campinchi R, et al: [Ophthalmological study of 6 cases of carotid-cavernous fistula treated with Fogarty catheter and embolization.] Rev Otoneuroophtaimol 50: , 1978 (Fre) 13. Fierstien SB, DeFeo D, Nutkiewicz A: Complete obliteration of a carotid cavernous fistula with sparing of the carotid blood flow using a detachable balloon catheter. Surg Neurol 9: , Gronski HW, Creely JJ Jr: Carotid-cavernous fistula: a complication of maxillofacial trauma. South Med J 68: , Hieshima GB, Mehringer CM, Grinnell VS, et al: Emergency occlusive techniques. Surg Neurol 9: , 1978 J. Neurosurg. / Volume 55 / November,

15 G. Debrun, et al. 16. Hosobuchi Y: Electrothrombosis of carotid-cavernous fistula. J Neurosurg 42:76-85, Kerber C: Use of balloon catheters in the treatment of cranial arterial abnormalities. Stroke 11: , Laws ER Jr, Onofrio BM, Pearson BW, et al: Successful management of bilateral carotid-cavernous fistulae with a trans-sphenoidal approach. Neurosurgery 4: , Lepoire J, Picard L, Montaut J, et al: [Treatment of carotid-cavernous fistula by endo-arterial occlusion using a balloon catheter.] Neurochirurgie 20: , 1974 (Fre) 20. Manelfe C, Berenstein A: Treatment of carotid cavernous fistulas by venous approach. J Neuroradioi 7: 13-21, McCormick WF, Kelly PJ, Sarwar M: Fatal paradoxical muscle embolization in traumatic carotid-cavernous fistula repair. Case report. J Neurosurg 44: , Mullah S: Experiences with surgical thrombosis of intracranial berry aneurysms and carotid cavernous fistulas. J Neurosurg 41: , Mullan S, Duda EE, Patronas N J: Some examples of balloon technology in neurosurgery. J Neurosurg 52: , Parkinson D: Carotid cavernous fistula: direct repair with preservation of the carotid artery. Technical note. J Neurosurg 38:99-106, Peeters F: [The technique of inflating and releasing balloons for the management of carotid cavernous fistulas.] ROEFO 129: , 1978 (Ger) 26. Peterson E, Valberg JD, Ventureyra E: Percutaneous treatment of carotid cavernous ftstula with preservation of the carotid artery. Presented at the Canadian Meeting of Neurological Sciences, June 19, Picard L, Lepoire J, Montaut J, et al: Endarterial occlusion of carotid-cavernous sinus fistulas using a balloon tipped catheter. Neuroradiology 8:5-10, Prolo D J, Burres KP, Hanbery JW: Balloon occlusion of carotid-cavernous fistula: introduction of a new catheter. Surg Neural 7: , Scott TE Jr, Tweed CG, Hollis BF, et al: Management of traumatic carotid cavernous fistula using the Fogarty catheter technique. J Trauma 20: , Sekhar LN, Heros RC, Kerber CW: Carotid-cavernous fistula following percutaneous retrogasserian procedures. Report of two cases. J Neurosurg 51: , Serbinenko FA: Balloon catheterization and occlusion of major cerebral vessels. J Neurosurg 41: , Serbinenko FA: Six hundred endovascular neurosurgical procedures in vascular pathology. A ten-year experience. Acta Neurochir (Suppl 28): , Taki W, Handa H, Yamasata S, et al: Embolization and superselective angiography by means of balloon catheters. Surg Neurol 12:7-14, Wagner O, Wagner M, Kletter G, et al: [Carotid-cavernous sinus fistula following the use of a Fogarty catheter. An unusual complication during the management of an acute traumatic occlusion of the internal carotid artery. Vasa 7:77-81, 1978 (Get) 35. Wilson WB, Bringewald PR, Hosobuchi Y, et al: Transient third nerve palsy after electrometallicthrombosis of carotid cavernous fistulae. J Neurol Neurosurg Psychiatry 39: , 1976 Manuscript received April 7, Accepted in final form June 9, Address for Drs. Lacour and Caron: Hbpital Universitaire Henri Mondor, Paris, Cr&eil, France. Address reprint requests to: G6rard Debrnn, M.D., Department of Radiology, Massachusetts General Hospital, Boston, Massachusetts J. Neurosurg. / Volume 55 / November, 1981