Although it is clear from the neurosurgical literature that the mortality associated with cavernous sinus surgery is “acceptable,” it has been far less clear whether these procedures have resulted in ophthalmic benefit. Most series refer to improvement without any quantitative assessment. In 1965 Dwight Parkinson¹ directly approached a vascular lesion within the cavernous sinus in a procedure requiring hypothermia and partial circulatory arrest. This was the beginning of the modern era in cavernous sinus surgery, the last area of uncharted territory in neurosurgery. Literally hundreds of articles have been published since then, detailing surgical approaches to various lesions in and around the cavernous sinus. The ability to operate successfully within this area is largely predicated on an improved understanding of the anatomy and pathology in the parasellar region and the development of modern neuroimaging. Cavernous sinus pathology is not infrequently revealed on imaging studies obtained for other unrelated reasons. The majority of patients, however, are still diagnosed because of ophthalmic complaints, including decreased and double vision.

Although the survival rate in cavernous sinus surgery has been generally good, there has been reported morbidity, including damage to the second, third, fourth, fifth, and sixth cranial nerves, all of which have a significant impact on visual function. It is somewhat surprising that little ophthalmic attention has been paid to this recent tendency toward more aggressive cavernous sinus surgery. This study was undertaken to determine the safety of cavernous sinus surgery by prospectively following a series of patients undergoing aggressive cavernous sinus surgery. Evaluations of function were designed to be done as quantitatively as possible in the hope of detecting any improvement or worsening of visual function that could be attributed to the surgery. It was hoped that the data would help both patients and surgeons to devise appropriate treatment strategies.

In essence, this study seeks to record the natural history of cavernous sinus surgery. A single surgeon, one institution, and an ophthalmological follow-up provide a unique database to address the following issues:

• What is the morbidity and mortality associated with cavernous sinus surgery?
• What are the ophthalmological sequelae?
• Are these complications of sufficient magnitude to recommend that cavernous sinus surgery not be done?
• The data are dependent on the one surgeon and his technique. Might a different surgeon and technique produce better results?
• What are the alternatives to cavernous sinus surgery?

When Ridley⁵ suggested in 1695 that the cavernous carotid lay against the lateral wall of the sinus, he laid the foundation for one of the most contentious debates in central nervous system anatomy. The carotid was not put back in its appropriate place until 1932, when Weizenhoffer⁶ recognized the normal separation between the cranial nerves and the carotid artery. As recently as 1966, Bedford⁷ (after dissecting 34 cavernous sinus specimens) inaccurately described the carotid as directly applied to the lateral wall. One can speculate about the fixation status of his specimens. More recent studies have concentrated on the various venous spaces, lateral, and anterior to and occasionally around, the carotid artery.^8–15

The major controversy that persists to this day concerns the microscopic anatomy of the venous channels themselves. By his choice of a name, Winslow had assumed a trabeculated venous space. These trabeculations were present in an illustration published by Duke-Elder¹⁶ and in many earlier anatomy texts. Their presence was thought to be responsible for the frequency of thrombosis within the cavernous sinus. Bedford⁷ initially challenged the presence of trabeculations, suggesting that the sinus was largely free of obstruction. In 1949,¹⁷ and later in the 1980s,^18,19 Taptas suggested that the area was not an open venous sinus at all, but rather an irregular network of veins. Bonnet,²⁰ using microdissection and corrosion casts, supported this plexus theory, which was subsequently championed by Parkinson.^21–23 Krivosic,²⁴ Rhoton and associates,²⁵ and Hakuba and coworkers²⁶ have all suggested that there are probably anatomic features of both; that is, trabeculated venous spaces and various venous channels intermixed within the extradural parasellar space. The lateral wall, in particular, may contain venous channels.¹³

The earliest described pathology of the cavernous sinus relates to its vascular origin. Biumi published a description of a carotid cavernous aneurysm in 1765, and Blane described the results of a postmortem on a woman who had died in 1794 with an intracavernous carotid aneurysm. Adams²⁷ recognized an aneurysm of the cavernous carotid associated with complete ophthalmoplegia and numbness. In his series of intracerebral aneurysms, Bartholow²⁸ noted a case of a carotid cavernous aneurysm in 1872.

Although various (usually fatal) attempts at draining the cavernous sinus occurred in the 19th century, Krogius²⁹ is generally credited with the first surgical approach to a “mesothelioma” (likely a meningioma) invading the cavernous sinus. Frazier³⁰ related the details of a case of cavernous sinus thrombosis, which had been treated surgically in 1900, concluding that “the cavernous sinus is not within the realm of the surgeon’s knife.” Langworthy³¹ suggested that the treatment of cavernous sinus thrombosis “consists in incising and draining the cavernous sinus directly.” Prior unsuccessful attempts are also mentioned. He does not detail whether he personally witnessed a successful procedure.

Prior to Parkinson’s report in 1965, there was little enthusiasm for surgery in this area. As late as 1978, Trobe and coworkers³² reviewed a series of 6 cavernous meningiomas and 9 aneurysms and concluded that “craniotomy is not recommended.” In 1979 J. Lawton Smith³³ commented that “neither of the two lesions [meningioma or aneurysm] in the cavernous sinus should be considered surgical candidates.” At almost the same time, Vinko Dolenc,³⁴ after careful anatomical studies, undertook a direct surgical attack on intracarotid vascular lesions in 7 patients, without resorting to modification of the circulation. These cases, predicated on Parkinson’s pioneering work, were the opening salvo in what has become a barrage of surgical cases involving the cavernous sinus.

Anatomically,⁴² the cavernous sinus can be seen as an extradural parasellar extension of the contents of the orbit.⁴³ The medial wall of the cavernous sinus directly continues as the medial periorbita, through the superior orbital fissure. The lateral wall of the cavernous sinus is formed by the condensation of the sheaths of the third, fourth, and first division of the fifth cranial nerves. This sheath condensation extends anteriorly to join the orbit at the superior orbital fissure, where the lateral wall fuses with the lateral periorbita. Kehrli and colleagues^9–11 have emphasized the true extradural nature of the cavernous sinus in a series of articles.

The bones adjacent to the cavernous sinus include the body of the sella (medially) and the floor of the middle cranial fossa (inferiorly). The floor of the cavernous sinus extends between the foramen rotundum (anteriorly) and the foramen ovale (posteriorly and laterally). The carotid artery enters the posterior aspect of the cavernous sinus through the floor overlying the foramen lacerum. The anterior clinoid, which represents the terminal portion of the lesser wing of the sphenoid, forms the anterior aspect of the lateral wall of the optic canal. Its underlying strut separates the superior orbital fissure from the optic canal. This potential space can be made into a working space by the neurosurgeon. Removal of the anterior clinoid reveals the anterosuperior aspect of the cavernous sinus, and access is provided via the roof of the cavernous sinus through the (anteromedial) triangle formed by the optic nerve (medially) and the third nerve (laterally).⁴⁴ Surgical approach through this space is a particularly convenient route to the anterior intracavernous carotid artery (see below).

The dural roof of the cavernous sinus continues¹⁴ as the diaphragma sellae, which covers the sella turcica enclosing the pituitary gland. The roof of the cavernous sinus extends posteriorly to the clival dura and laterally to the posterior clinoids. The posterior clinoids also mark the rostral termination of the dorsum sellae. The base of the posterior wall of the cavernous sinus is marked by the petroclinoid (Gruber’s) ligament, which connects the posterior clinoid to the petrous apex.⁴⁵ This dural attachment forms the roof of Dorello’s canal, which contains the sixth nerve after it enters the dura, and the inferior petrosal sinus on its way to the jugular bulb.

The carotid artery, the largest structure contained within the cavernous sinus, enters the sinus from the carotid canal, which is located within the petrous bone, where the carotid runs just under the greater superficial petrosal nerve and above the Eustachian tube. At the medial end of the carotid canal, it makes a turn (Dolenc’s lateral loop)⁴⁴ to rise over the area of the foramen lacerum. It enters the floor of the cavernous sinus just medial to Meckel’s cave and makes another loop (Dolenc’s medial loop) to travel forward in its intracavernous horizontal section. As mentioned, the carotid is usually in close contact with the lateral sella wall and thus the medial wall of the cavernous sinus. A condensation of fibrous tissue separates it from the sella. At the anterior end of the horizontal section, the carotid artery makes its final loop (anterior), reversing its course just under the anterior clinoid and medial to the optic nerve. Two condensations of fibrous tissue, the proximal and distal loops, surround the carotid below and above the clinoid and mark the passage of the carotid, first out of the cavernous sinus, and then into the intradural space.^46–48 Thus the clinoidal portion of the carotid artery is still extradural.

The lateral wall of the cavernous sinus³⁷ contains the third nerve superiorly, the fourth nerve just below it, and the first division (the ophthalmic) of the fifth nerve. The second division of the fifth nerve (maxillary) forms the inferior lateral border of the cavernous sinus and parallels the first division. The maxillary division runs through the foramen rotundum located just below the inferior aspect of the superior orbital fissure, which transmits the ophthalmic division. As the cranial nerves approach the superior orbital fissure, the third nerve divides into a superior and inferior division. The fourth nerve travels superiorly to cross laterally to medially above the branches of the third nerve. The ophthalmic division of the fifth nerve divides into 3 branches: the lacrimal, frontal, and nasociliary. The blood supply to the cranial nerves within the cavernous sinus has been studied in detail.^49–56

Textbooks older than 50 years often list the ophthalmic artery as the first major branch of the intracranial internal carotid artery. Actually, 2 variable sets of carotid branches exit in and around the cavernous sinus.^57,58 They are particularly important because they supply blood to the surrounding dura and to the cranial nerves running in and around the cavernous sinus. The first of these branches, the meningohypophyseal trunk,⁵⁶ divides into the tentorial artery, the dorsal meningeal artery, and the inferior hypophyseal artery. The tentorial artery supplies the cranial nerves as they enter the cavernous sinus posteriorly. The second major vessel is the inferolateral trunk,^51,52,55 which subsequently divides into 4 branches. The superior 2 branches are critical to the blood supply to the intracavernous third and fourth cranial nerves. Significant collateralization in this area usually protects these nerves, even if the inferolateral trunk is disturbed. Terminal dural branches from the middle meningeal artery (entering the skull through the foramen spinosum) connect with the branches of the inferolateral trunk to form the artery of the foramen rotundum (accompanying the second division of cranial nerve V) and the artery of the foramen ovale (accompanying the third division of the trigeminal nerve). The inferolateral trunk is variable in location, but is found in the majority of specimens studied. It may rarely arise as a branch of the meningohypophyseal artery at the posterior aspect of the cavernous sinus. There are additional collaterals that join the inferolateral trunk to the accessory meningeal branches that supply the pterygoids and the inferotemporal area below the middle cranial fossa.

Besides the venous channel, the cavernous sinus also contains the sixth nerve, which is located just lateral to the carotid artery, and the sympathetic plexus, which enters the cavernous sinus within the carotid sheath.^59–64 The pericarotid sympathetic plexus coalesces to form a variable number of trunks, which exit with the sixth nerve to join branches of the fifth nerve entering the superior orbital fissure. In addition, due to its embryological and anatomical connection to the orbit, it is not surprising that fat can be found^10,65 microscopically in 10% or more cases. This finding has produced some confusion because the low signal intensity within the cavernous sinus on computed tomography has suggested a pathological process in patients with other surrounding pathology.

The venous connections into and out of the cavernous sinus include the superior ophthalmic vein anteriorly and various cortical branches that may enter the cavernous sinus directly. In addition, in a minority of cases, the inferior orbital vein may directly enter the cavernous sinus, instead of indirectly via its connection to the superior ophthalmic vein anterior to the superior orbital fissure. Outflow from the cavernous sinus usually proceeds via the inferior petrosal sinus, which exits the posterior aspect of the cavernous sinus and runs directly down to the jugular bulb or through the superior petrosal sinus to the lateral sinus. In addition, there are variable connections to the pterygoid inferiorly through a plexus of veins and to the opposite cavernous sinus through an intrasellar connection both anterior and posterior to the pituitary fossa.⁶⁶ Finally, there may be a variable venous plexus that extends down the clivus posteriorly and under the dural aspect of the middle cranial fossa inferiorly and laterally.

Because there are no valves within the cavernous sinus, blood flow can easily be reversed, particularly in the setting of arterial injection into the cavernous sinus either directly from the carotid artery or indirectly through internal or external dural branches. In the setting of a carotid cavernous fistula, flow often reverses into the superior ophthalmic vein, thereby producing evidence of orbital congestion as well as secondary increased intraocular pressure (related to the problems with ocular venous outflow). This flow reversal results in the myriad ophthalmic manifestations of a carotid cavernous fistula.^67–71

Although the cavernous sinus is separated from the afferent visual pathway by the optic strut anteriorly and by the increasing subarachnoid separation between the optic nerve and the roof of the cavernous sinus posteriorly, pathology arising in the cavernous sinus can extend superiorly to affect the optic nerve or chiasm. Often extracavernous extension elevates the optic nerve, compressing it against the falciform fold in the dura at the posterosuperior exit of the optic canal.⁷² This results in arcuate (usually inferior) visual field defects and variable acuity loss.

A second exception was cavernous sinus thrombosis. In 1854, MacKenzie,⁷³ in his 4th edition, reported on a case of a Welsh laborer who was struck in the orbit with a clay pipe and who eventually succumbed to a cavernous sinus infection. At autopsy a retained foreign body was found within the cavernous sinus. Much of the early ophthalmic literature that dealt with the cavernous sinus reported cases of septic cavernous sinus thrombosis. This was usually secondary to head and neck infectious sources commonly originating in the sinuses, teeth (dental caries), and the ears (mastoiditis and petrositis).⁷⁴ The prognosis in pre-antibiotic days was dismal, since death almost always occurred within days; Frazier³⁰ estimated that only 7% survived.

Medical science was slow to recognize that mass lesions in the parasellar region could produce cranial nerve palsies. In the 19th century, most cranial nerve palsies were attributed to inflammatory, toxic, or metabolic pathology that was thought to be intra-axial. Infectious processes, including syphilis and tuberculosis, were frequently blamed for the development of ophthalmoplegia. Gowers⁷⁵ recognized that most intra-axial pathology usually affected the ocular motor nerves bilaterally. He did, however, note that “paralysis of the ocular muscles may be due to disease of the nerves in the orbit or at the base of the brain.” Swanzy,⁷⁴ writing in Norris and Oliver, noted that “tumors and inflammatory products about the cavernous sinus are very liable to cause partial or complete third-nerve paralysis, along with partial or complete paralysis of some or all of the other orbital nerves, as well as of the optic and olfactory nerves.” Langworthy³¹ (in Casey Wood’s The American Encyclopedia of Ophthalmology) noted that “from time to time observers report neoplasms involving the cavernous sinus in which the eye symptoms vary according to the extent and character of the growths.” Nettleship, quoted by Swanzy,⁷⁴ reported a “sarcoma” that occupied the right cavernous sinus resulting in total ophthalmoplegia. He recognized that despite mild proptosis, the absence of optic nerve dysfunction suggested sparing of the orbital apex.

Wilbrand and Saenger⁷⁶ summarized previously reported cases of skull base pathology resulting in ocular motor problems in a text published at the turn of the century. Frank Walsh⁷⁷ discussed cavernous sinus thrombosis and carotid cavernous fistulae, but in the first edition of his text on clinical neuro-ophthalmology had little to say about cavernous sinus neoplasms. Although Duke-Elder¹⁶ showed a picture of a patient with ptosis and probable ophthalmoplegia (captioned as a “para-orbital tumor”), he failed to discuss the possibility of neoplastic lesions of the cavernous sinus.

Neoplastic pathology was often overlooked as a cause of neurogenic ophthalmoplegia. Duke-Elder⁷⁸ listed tumors as only one of 6 possible etiologies (including acute and subacute inflammatory diseases, metabolic diseases, intoxications from exogenous poisons, vascular lesions, and trauma). In his list of the causes of chronic and progressive ophthalmoplegia (in which he included syphilis, multiple sclerosis, diffuse sclerosis, syringomyelia, and amyotrophic lateral sclerosis), he omitted mass lesions entirely. In localizing intracranial tumors, he mentions the brainstem, pons, medulla, supratentorial regions (causing herniation), and the meninges, but without discussing the parasellar region.

In view of the fact that meningiomas make up the majority of cavernous sinus lesions, it is remarkable that Cushing and Eisenhardt⁷⁹ did not identify the cavernous sinus as a significant primary location for meningiomas. Only 5 tumors were identified as primarily parasellar in the 294 patients they studied. They did recognize that tumors originating in Meckel’s cave, the medial sphenoid wing (including the anterior clinoid), and the floor of the middle cranial fossa could secondarily involve the tissues of the cavernous sinus. Undoubtedly, extension into the cavernous sinus was responsible for some of the impairment of the second through sixth cranial nerves noted in many of their patients.

Individual case reports appeared during the first 2 decades of the 20th century,⁸⁰ but it was Foix⁸¹ who first reported on a series of cases with cavernous sinus pathology. This pioneering work was followed by a 1938 report from Geoffery Jefferson⁸² on 55 saccular aneurysms of the cavernous sinus. Jefferson also recognized the importance of these findings to ophthalmologists, and when he delivered the 1952 Bowman lecture,⁸³ he described the features of 112 cavernous sinus lesions (including 22 traumatic cases with 17 carotid cavernous fistulae, 38 aneurysms, and 52 tumors). He reported that there was “little in the structure of the cavernous sinus to furnish material for primary neoplasms.” Thus, the majority of tumors in his series were related to extrinsic invasion, most commonly from the nasopharynx or paranasal sinuses.

Current data would suggest that he was correct about truly primary tumors of the cavernous sinus. Only aneurysms are literally primary within the cavernous sinus.^84,85 Although meningiomas can arise from the dura of the lateral wall of the sinus, most originate from dura covering the sphenoid or petrous bones and secondarily invade the cavernous sinus. These tumors have a propensity for following the sheaths of the cranial nerves as they enter the lateral wall of the sinus.³⁹ Jefferson listed 3 meningiomas (out of 346 meningiomas that he had treated) that were located primarily within the cavernous sinus. He also included 4 “neurinomas” involving Meckel’s cave and one case of neurofibromatosis. The majority of his cavernous sinus cases was malignant and included 23 nasopharyngeal carcinomas, 8 carcinomatous metastases, 2 adamantinomas, and 1 chordoma.

Of 102 patients from the Mayo Clinic reported on in 1970,⁸⁶ tumors occurred in 70 (69%), aneurysms or fistulas in 19 (19%), and inflammation in 9 (9%). Forty-three (61.4%) of the 70 patients with neoplastic disease suffered from metastasis. This included distant spread in 23 patients and local spread (nasopharyngeal carcinoma in all but one) in 20. Only 14 patients (20% of those neoplastic lesions) had primary intracranial lesions including 6 pituitary adenomas, 3 meningiomas, 2 craniopharyngiomas, 2 sarcomas, and 1 neurofibroma.

More recently (in 1996), James Keane⁸⁷ published an epidemiological study reviewing 151 patients who presented with the “cavernous sinus syndrome” during a 26-year period. Inclusion criteria for the study included evidence of multiple cranial nerve palsies. This was also a very select series, because these individuals were evaluated while they were patients at the Los Angeles County Hospital. It is thus not surprising that trauma occurred in 36 patients (24%). An additional 17 patients (11%) had conditions secondary to surgical trauma following neurosurgical procedures. This high frequency of trauma diluted the percentage of tumors (45 [30%]) and aneurysms (9 [6%]). Nineteen patients (13%) had evidence to suggest inflammation and an additional 15 patients (10%) were thought to have “likely” inflammation. The other interesting feature in this series is that although the incidence of nasopharyngeal carcinoma had declined (from 46% in the Jefferson series to 22% in the Keane series), it still was the single most frequently occurring tumor. In addition, metastatic disease occurred in 8 patients and lymphoma in an additional eight patients (18% each). Invasive pituitary adenomas were recognized in 8 patients (18%) and meningiomas in only 4 patients (9%).

The major difference between the previous studies and this most current publication has been the advent of neuroimaging. Introduced in the United States in 1975, computed tomography^88–90 and, more recently, magnetic resonance imaging^91–96 have completely revolutionized our ability to detect lesions in the cavernous sinus.⁹⁷ Prior to that, cavernous sinus pathology was detected almost solely on the basis of clinical findings, most commonly diplopia (due to involvement of the third, fourth, and sixth cranial nerves). Less commonly, patients were evaluated for numbness or pain. Prior to 1975 it was rare for cavernous sinus pathology to be recognized without evidence of progressive cranial nerve dysfunction.

Recent larger surgical series provide additional epidemiological data (see Appendix B). In a series of 63 cases treated by Dolenc and colleagues⁹⁸ between 1980 and 1985, there were 40 cases of meningiomas, 2 cases of malignant meningiomas, 7 pituitary tumors, 4 neurilemomas, and 3 plexiform neurofibromas. Other pathology included fibrous dysplasia, epidermoid, cholesteatoma, myxoma, and fibrosarcoma. In 154 cases collected from 3 institutions, Al-Mefty and coworkers⁹⁹ reported 42 meningiomas, 35 pituitary tumors, 32 aneurysms (including 19 ophthalmic and 13 carotid), and 4 fistulae. He also listed 15 nasopharyngeal carcinomas and 12 other malignant tumors, including 4 metastases, 4 paranasal sinus carcinomas, 2 adenoid cystic carcinomas, and 2 myxochondroid sarcomas. This increased representation of meningiomas (and benign tumors in general) may have a component of referral bias.

The proliferation of reported cavernous sinus pathology is indicative of the substantial improvement in our diagnostic techniques, especially neuroimaging, rather than an actual increase in the frequency of these lesions. This interpretation is supported by the decreasing incidence of diplopia and decreased vision as primary symptoms in referred patients and increasing frequency of pain and headache. In many of these patients, the presence of cavernous sinus pathology was uncovered fortuitously, when imaging studies were ordered for nonspecific headache syndromes or migraine. Without numbness, pain has a low probability of indicating a cavernous sinus lesion.

In a retrospective epidemiological study (Appendix A), files of patients seen in the neuro-ophthalmology unit at the University of Virginia were screened for cavernous sinus involvement. A total of 347 patients (230 women and 117 men) seen during the last 12 years were selected as having been coded for cavernous sinus pathology. Neoplasia affected the largest group, that is, 236 (63%) of 374 cases. Of the tumorous growths, meningiomas (118) made up one-half with a smaller contribution of 35 pituitary tumors of 12 neurilemomas, and 6 cavernous hemangiomas. Pituitary tumors are rarely reported within the cavernous sinus.^100,101 Carcinomas were much less common (only 25 cases) and were scattered among the various types. The other major diagnostic group consisted of vascular lesions (97 cases), including 45 aneurysms and 52 fistulae, both direct and indirect. Inflammatory lesions were generally uncommon; 4 cases of cavernous sinus thrombosis, 7 presumed Tolosa-Hunt syndrome (one with biopsy confirmation), 2 Zoster inflammation, 2 granulomatous disease (one case related to sarcoid), 3 aspergillomas, and 2 mucormycosis. Patient ages ranged from 5 to 84 years, with a mean of 51 years at presentation. In analyzing the larger diagnostic groups separately, only neurilemomas occurred in patients at a significantly younger age (31 years) than the overall population.

Meningiomas are the most frequently occurring tumor in all modern lists. Although rare cases of entirely intracavernous meningiomas occur, the majority (as suggested by Cushing) arise from the surrounding dura (Figure 1). They may originate from the anterior clinoid.^99,102 Delfini and colleagues¹⁰³ reported on 16 patients with meningiomas arising from Meckel’s cave. These patients had usually presented with symptoms of trigeminal dysfunction. Clival meningiomas can also invade the cavernous sinus, in which case they commonly produce sixth cranial nerve dysfunction given that Dorello’s canal is affected.^39,104 Although clival meningiomas most frequently cause brainstem dysfunction (due to posterior extension), many expand anteriorly, affecting the posterior cavernous sinus. This involvement increases the morbidity associated with surgical approaches.¹⁰⁵ Petroclival meningiomas may also be very extensive and difficult to excise completely.

FIGURE 1 Case 2. Computed tomography (left) and magnetic resonance imaging (right) scans of a patient with a long history of a known skull base meningioma. The tumor can be seen to be encasing the carotid and middle cerebral arteries.

The initial series consisted of a total of 81 patients who underwent 82 cavernous sinus surgical procedures. Of these patients undergoing cavernous sinus surgery, 56 were found postoperatively to have meningiomas (Table 1). This series included 16 men and 40 women whose ages ranged from 23 to 81 years with a mean of 51 years. One man had a recurrence and underwent a second procedure to extirpate the cavernous sinus meningioma. The time from initial development of symptoms to diagnosis ranged from less than 1 week to 168 months, with a mean of 18.9 months. Diagnosis was delayed by more than 2 months in 23 of 57 cases (as much as 13 years). In one remarkable case the diagnosis was delayed for 2 years in spite of the fact that the patient had previously had a meningioma resected 10 years earlier. Diagnoses made prior to the correct identification included microvascular cranial nerve palsies in 3 patients. Sinus disease, “lazy eye,” and multiple sclerosis were each diagnosed in 2 patients. Other initial diagnoses included myasthenia, thyroid disease, labyrinthitis, trigeminal neuralgia, migraine, hypothyroidism, and “nerves.” Twenty-one patients (37%) had previous surgery.

TABLE 1 DETAILS OF PATIENTS UNDERGOING CAVERNOUS SINUS SURGERY

TABLE 2 SYMPTOMS OF CAVERNOUS SINUS PATHOLOGY IN CURRENT STUDY AND LITERATURE

Other surgeons routinely remove the zygomatic arch, thus making this approach into an orbitozygomatic pterional one.^102,113,114 This was believed to be unnecessary in this series. A single burr hole was placed just behind the superior portion of the lateral orbital rim. This ideally entered the orbit as well as the anterior cranial fossa. A second burr hole was placed posteriorly in the temporal region. The dura was freed from the overlying bone, and a pterional-based bone flap was raised, which permitted immediate view of the dura of both the anterior and middle cranial fossae.

An extradural dissection was then carried down to the foramen spinosum, where the middle meningeal artery was identified and ligated. Dissection was carried forward, while identifying the foramen ovale and the foramen rotundum anteriorly, all extradurally. The course of the intrapetrous carotid artery was also identified. The carotid canal,¹¹⁵ within the petrous bone, was opened by dividing the greater superficial petrosal nerve (to prevent traction on the facial nerve) and by carefully removing bone back to the posterior loop of the carotid artery. Other surgeons have used this portion of the carotid artery to base a bypass graft when carotid sacrifice is planned.^116–118

One of the disadvantages of this approach is the loss of the greater superficial petrosal nerve, which results in decreased reflex tearing by interrupting the parasympathetic innervation to the lacrimal gland. Wright¹¹⁹ has more recently advocated a longer saphenous vein bypass graft from the cervical carotid to the distal supraclinoid carotid to avoid sacrificing the greater superficial petrosal nerve. Other surgeons believe that the possible complications of carotid revascularization procedures (as part of cavernous sinus surgery) outweigh their potential usefulness in increasing resectability.¹²⁰ De Monte¹²¹ has written that carotid sacrifice is rarely, if ever, indicated. In this series, although 1 carotid artery ruptured and required sacrifice and 2 carotids were oversewn, no attempts at revascularization were made.

The periorbita and the dura are dissected free from the orbital roof, which is then removed. The greater wing of the sphenoid (the lateral wall of the orbit) is also taken down with ronguers, back to the superior orbital fissure and the foramen rotundum (Figure 2). The anterior clinoid, which represents the terminal portion of the lesser wing of the sphenoid, is carefully hollowed out by using a high-speed drill with a diamond burr. The residual bony rim is removed with curettes. In a small percentage of patients, the anterior clinoid may be connected via a bony strut back to the middle clinoid,⁴³ which makes removal of the anterior clinoid difficult and potentially more dangerous to the surrounding structures, especially the underlying optic nerve and carotid artery. The optic strut, which separates the optic canal from the superior orbital fissure, is further removed with the aid of rongeurs and the drill. Because there is aeration of the anterior clinoids in a small percentage of patients, an opening can be made into the paranasal sinuses. To avoid cerebrospinal fluid (CSF) leak, it is imperative to completely close the potential opening, usually with muscle and fibrin glue. Despite these precautions, CSF leaks are common in cavernous sinus skull base surgery.

FIGURE 2 Case 30. Postoperative computed tomography scan following a pterional approach to the cavernous sinus. The roof and lateral wall, including the anterior clinoid, are gone on the left side.

Once the extradural portion of the operation is complete, the dura is opened. In the early cases in this series, the dura was opened inferiorly, thereby leaving a sheet of dura to protect the temporal lobe. A malleable retractor was then used to elevate the temporal lobe posteriorly and superiorly. Several of the early patients suffered detectable damage to the temporal lobe (infarct and secondary encephalomalacia) from long-duration compression by the static retractor. In later cases, a more medial approach was used by opening the Sylvian fissure, which permitted access to the lateral wall of the cavernous sinus without putting traction on the temporal lobe. In one case of a cavernoma, the cavernous sinus was entered without ever opening the dura by splitting the superior orbital fissure (following bony excision).

The direction of approach to the lesion within the cavernous sinus depends on the location of the tumor and its extent. In larger tumors that have secondarily invaded the cavernous sinus, often the tumor becomes obvious within the area of the middle cranial fossa or extends above the area of the cavernous sinus. When the tumor is small and localized, the cranial nerves, found within the lateral wall of the cavernous sinus, must be identified to prevent direct damage. This is usually relatively easy with small lesions, but it may be more difficult when the tumors are larger and especially when they are infiltrative.

The earliest surgical opening into the cavernous sinus (in the modern era of skull base surgery) was via the potential space between the fourth cranial nerve above and the fifth cranial nerve below.¹ This triangular space has become known as Parkinson’s triangle in honor of Dwight Parkinson, who initially described this surgical approach. The spaces between each of the cranial nerves also offer potential access to lesions within the cavernous sinus.⁴⁴ Another common approach is through the anterior cavernous roof. By removing the anterior clinoid, the space between the optic nerve and oculomotor nerve (the anteromedial triangle) opens directly into the anterior medial cavernous sinus.

Tumors often splay the cranial nerves apart, pointing the surgeon in the direction of easiest access. The various approaches have been enumerated by Dolenc in a series of named triangles. Surgery through each may be tailored to particular lesions, depending on their location. Other less common approaches to the cavernous sinus include a direct subtemporal approach to the lateral wall,¹²² transmaxillary¹²³ or transfacial¹²⁴ access, and transphenoidal¹²⁵ and transnasal approaches.¹²⁶ Several of the latter involve endoscopy.¹²⁷ Lesions that extend into the posterior cranial fossa may require more extensive surgery (beyond the scope of this review). None of our patients required these alternative procedures.

One of the factors that prevented earlier surgical approach to cavernous sinus lesions was the fear of causing uncontrollable bleeding. The earliest surgery within the cavernous sinus was performed with patients under hypothermia and hypotensive anesthesia or even in cardiac arrest. Although bleeding still may present a challenge during surgery, this issue has turned out to be less of a problem than otherwise expected, given that the lesion itself displaces the venous space. Although there may be bleeding from the tumor, cavernous sinus bleeding per se is often not problematic until the tumor boundaries are reached. Surgical packing of the residual venous spaces usually controls blood within the surgical field. Several tumors in this series had residual bleeding despite the preoperative embolization, which was routinely performed. All of these instances were controlled intraoperatively.

TABLE 3 PREOPERATIVE OPTIC NEUROPATHY IN PATIENTS UNDERGOING CAVERNOUS SINUS SURGERY

Five (25%) of the 20 patients showed immediate improvement, 8 (40%) experienced no change, and in 6 (30%) the dysfunction was worse postoperatively. The patient with NLP vision experienced no change. The patient with preoperative 20/400 vision could not be reassessed before leaving the hospital. Ten of the 20 patients with preoperative optic neuropathies underwent long-term follow-up (3 months or longer). Two of these (20%) had better central acuity than preoperatively, the neuropathy in 4 (40%) was the same, and 4 (40%) were worse than before surgery. Five patients (of 19 with preoperative optic neuropathies) had NLP (2 with short-term follow-up only), and the one patient preoperatively with NLP remained so. This was often anticipated preoperatively because tumor encased the optic nerve.

Thirty-seven patients had no evidence of preoperative optic nerve dysfunction. Thirty-three (88%) remained free of optic nerve dysfunction postoperatively, but four (11%) developed evidence of such dysfunction (Table 4). All 4 patients developed partial optic neuropathies with evidence of an afferent pupillary defect and visual field defects present on automated static perimetry (Figure 3). The field defects were characteristically arcuate in nature. There was relatively good preservation of central visual function in all 4 patients with new optic nerve dysfunction. There was marginally greater risk of a new optic neuropathy if the patient had previous surgery (2 of 14 [14%] without prior optic nerve damage) than if the patient had not had previous surgery (2 of 23 [8.7%]).

TABLE 4 ACQUIRED OPTIC NERVE DAMAGE FOLLOWING CAVERNOUS SINUS SURGERY

FIGURE 3 Case 6. Visual fields, left eye (left) and right eye (right), following resection of cavernous sinus tumor. In 1986, a 36-year-old patient was referred with intermittent double vision. At that time, he had evidence of a minimal abduction deficit OS, and (more ...)

TABLE 5 PREOPERATIVE SIXTH NERVE DYSFUNCTION IN PATIENTS UNDERGOING CAVERNOUS SINUS SURGERY

Long-term follow-up data were available in 16 of the original group of 24 patients with preexisting sixth nerve palsies. Seven remained complete, 7 were incomplete, and 2 recovered (Table 5). Neither of the patients who had long-term clearing of prior sixth nerve dysfunction had undergone transcranial surgery. Of the 6 patients with complete palsies preoperatively, long-term follow-up data were available in 4 cases, none of whom cleared.

Thirty-three patients had no evidence of an abduction deficit preoperatively. Twenty-five (76%) developed new sixth nerve palsies postoperatively (Table 6). Only 8 patients retained complete abduction immediately postoperatively. Only 2 (17%) of 12 patients who had previous transcranial surgery but no sixth nerve palsy preoperatively remained free of sixth nerve dysfunction, whereas 6 of 21 without prior surgery (29%) managed to have no evidence of abduction deficit postoperatively. Acute abduction deficits often improved. Only one patient with a surgically acquired sixth nerve palsy had no return of function with long-term follow-up. Seven patients with sixth nerve dysfunction preoperatively remained complete with long-term follow-up, including 3 whose defect was incomplete preoperatively. Seven of 25 patients with abducens palsies postoperatively recovered completely, but 6 retained incomplete abduction deficits. Five of 7 patients who recovered had experienced a complete deficit immediately postoperatively, and 2 had experienced an incomplete deficit. Combining the 2 groups at their last follow-up visit, only 9 (30%) of the 30 patients with long-term follow up were completely free of sixth nerve dysfunction, 8 (27%) had complete abduction defects, and 13 (43%) had partial dysfunction.

TABLE 6 NEW SIXTH NERVE DYSFUNCTION FOLLOWING CAVERNOUS SINUS SURGERY

TABLE 7 PREOPERATIVE THIRD NERVE DYSFUNCTION IN PATIENTS UNDERGOING CAVERNOUS SINUS SURGERY

None of the 19 patients with preoperative third nerve palsies recovered immediately following surgery. The one patient with a complete third nerve palsy preoperatively remained so postoperatively, and an additional 15 patients progressed from incomplete to complete oculomotor paralysis. Prior transcranial surgery slightly increased the risk of progression from incomplete to complete. Five of 6 patients who had prior surgery and incomplete third nerve palsies progressed (83%), whereas 77% of patients without prior surgery had their third nerve palsy become complete (10 of 13). In the short-term postoperative analysis, 7 continued to have no evidence of pupillary involvement. An additional 6 (60%) of the previous 10 patients with no pupil involvement developed evidence of pupillary dysfunction.

Variable improvement tended to occur. We had long-term follow-up data in 12 of the 19 patients who had had preoperative third nerve palsies. None of these resolved completely, 8 partially resolved, and 4 had persistent complete third nerve paralysis. Three patients who had only experienced incomplete dysfunction preoperatively had complete third nerve palsies in the long term. No patients with preexisting oculomotor dysfunction showed improvement in their third nerve dysfunction There was evidence of pupillary dysfunction in 9 of the 12 patients with residual third nerve dysfunction, despite the partial recovery in the long term. In addition, 2 of the 8 patients with residual incomplete third nerve palsies developed aberrant regeneration. Both of these patients had undergone prior intracranial surgery.

Thirty-seven (97%) of the 38 patients who did not have a third nerve palsy preoperatively developed evidence of third nerve dysfunction postoperatively (Table 8). Only one patient with a meningioma postoperatively remained completely free of third nerve dysfunction. Of these new cases of third nerve dysfunction, 25 (68%) were complete and 12 (32%) were incomplete. Pupillary involvement broke down almost equally, with evidence of pupillary dysfunction in 19 but none in 18. There were long-term follow-up data in 23 patients in this subset. Thirteen cases (57%) cleared completely (Figure 4), and 10 (43%) cleared incompletely.

TABLE 8 NEW THIRD NERVE DYSFUNCTION FOLLOWING CAVERNOUS SINUS SURGERY

FIGURE 4 Case 15. A 47-year-old patient presented with a 3-month history of pain involving her left cheek. An otolaryngologist thought it was a sinus infection. She then saw 2 oral surgeons, who informed her that her symptoms were caused by her gums. A neurologist (more ...)

As an indication of the meticulous aspects of the cavernous dissection, no patient without oculomotor palsy preoperatively was left with a complete third nerve palsy, but 7 patients (37%) with incomplete defects did have evidence of aberrant regeneration (Figure 5).

FIGURE 5 Case 52. A 51-year-old patient presented with a 1-year history of intermittent but progressive binocular diplopia. She noticed that this was worse on left gaze and was accompanied with gradually increasing pain on the left side of her face, involving (more ...)

A total of 9 patients had aberrant regeneration of the third nerve, 2 with prior oculomotor deficits and 7 who developed third nerve dysfunction postoperatively. Of the preexisting third nerve dysfunction, one case of aberrant regeneration developed in a patient with prior surgery, and the other had no prior history of intervention. Of the 7 cases in patients without a prior third dysfunction, 3 occurred in patients with prior surgery (3 of 13 [23%]) and 4 were in patients without surgery (4 of 24) [17%].

The chance of complete recovery was substantially better if the palsy was incomplete following surgery. Fifteen of 16 patients (94%) who had incomplete clearing immediately postoperatively and had long-term follow-up recovered. When the third nerve palsy was complete following surgery, only 6 (32%) of 19 cases recovered totally. No cases of new oculomotor dysfunction remained complete. Of the 10 cases improving but not clearing, 7 demonstrated evidence of aberrant regeneration. There did not seem to be an increased risk associated with prior surgery. Although 4 of 10 patients who did not clear completely had prior intracranial surgery, 6 did not, and 1 of the 6 patients who demonstrated complete clearing had prior surgery. Both patients who developed new complete third nerve palsies had not had previous surgery. Thus, although patients with immediate worsening of third nerve function postoperatively usually improved, this was not universal and definitely not always complete.

TABLE 9 PREOPERATIVE FIFTH NERVE DYSFUNCTION IN PATIENTS UNDERGOING CAVERNOUS SINUS SURGERY

Loss of corneal sensation was of considerable clinical significance. Only 1 of 7 preoperatively anesthetic patients partially recovered with long-term follow-up (representing functional improvement). Both patients with completely anesthetic corneas preoperatively developed neurotrophic keratitis. Four additional patients with partial anesthesia preoperatively developed neurotrophic keratitis after becoming completely anesthetic postoperatively.

Thirty-four (77%) of the 44 patients who did not have evidence of preoperative fifth nerve dysfunction developed evidence of postoperative hypesthesia (Table 10). The risk of developing a new trigeminal palsy was slightly greater if the patient had undergone previous transcranial surgery (87% [13 of 15] vs 72% [21/29]). Twenty-one of these newly affected patients (62%) were completely anesthetic, and 13 (38%) had decreased sensation. Long-term follow-up data in 19 of these 34 patients revealed recovery in 2, incomplete dysfunction in 7, and complete anesthesia in 10. Five of the 13 who acquired anesthesia improved to hypesthesia with long-term follow-up. Of the 2 who completely recovered, both were partially compromised postoperatively. Six (60%) of the 10 persistently anesthetic patients developed neurotrophic keratitis.

TABLE 10 NEW FIFTH NERVE DYSFUNCTION FOLLOWING CAVERNOUS SINUS SURGERY

A more interesting question involves those preoperatively normal patients who developed optic nerve dysfunction postoperatively. The mechanism is unclear. While it is possible vasospasm may play a role, the most likely culprit is direct damage at the time of the clinoid removal. In this series the clinoid was removed extradurally, which may increase the risk over an intradural approach. This has been noted, although not emphasized, in the literature. In a series of 40 patients undergoing clinoid removal, 31 of which were done extradurally, 3 patients suffered postoperative diminution of visual acuity or visual field.¹²⁸ None of the 9 patients in whom an intradural approach was used had similar complications.

It is presumed that surgical manipulation around the optic nerve was responsible for the development of the optic nerve dysfunction in the 4 patients who developed partial but incomplete optic nerve damage. In particular, even the careful attempt to core the anterior clinoid with a high-speed drill puts the optic nerve at risk. There can be tremendous heat transfer with the use of a diamond burr (used to decrease bleeding and limit kickback in this tight space). Dolenc⁴⁴ specifically stressed the importance of adequate irrigation during drilling to minimize heat transfer. It is possible that significant temperature elevation occurred, despite the best efforts at continuous irrigation.

The incidence of new optic nerve damage in this case (7%) is substantially higher than that in previously reported series of cavernous sinus cases undergoing surgery. The easiest explanation for this incidence (not previously reported) is that without quantitative assessment of optic nerve function, these cases would likely have been missed clinically. Although the afferent pupillary defect ranged from 0.9 to 2.4 log units, central acuity was no worse than 20/50 in the 4 patients. Without measuring the afferent pupillary defect or obtaining quantitative perimetry for evidence of arcuate visual field defects, this dysfunction would have been overlooked. Thus, more careful attention to optic nerve function postoperatively may well detect unrecognized optic neuropathy.

It has been suggested by neurosurgeons that an “easy muscle procedure will correct a patient with VI nerve palsy.” Although it is true that muscle surgery can be done to provide binocularity centrally and probably with at least some degree of eccentric gaze, patients with complete sixth nerve palsies are much more difficult to align, with a resultant much smaller binocular field.

A pupil-sparing third nerve palsy due to an aneurysm of the posterior communicating artery is extremely rare. With the slow progression of cavernous sinus pathology, sparing of the pupil is not unusual, as demonstrated in our preoperative evaluation. Less often, more rapidly growing parasellar lesions may also produce pupil-sparing third nerve dysfunction.¹³⁰ Previous compromise of the third nerve by tumor influences a patient’s prognosis. None of the 19 patients who had preoperative third nerve palsies experienced complete recovery or substantial improvement. Since one of the indications for intervention was a patient’s desire to eliminate diplopia, evidence of third nerve dysfunction was a very poor prognostic marker of complete recovery following cavernous sinus surgery.

Our data support the initial observation of Dolenc and colleagues,⁹⁸ who noted that although there was a substantial incidence of third nerve palsies postoperatively, most cleared. One hundred four (90.4%) of 115 patients with intracavernous aneurysms who had undergone surgery developed third nerve palsies; all but 7 (6.1%) recovered.¹³¹ All of our patients with incomplete third nerve palsies postoperatively (and without evidence of third nerve dysfunction preoperatively) recovered completely. On the other hand, there are long-term follow-up data on 30 of the 40 patients who had complete postoperative third nerve palsies and only 8 cases (27%) cleared completely. Another 4 patients (13%) failed to recover at all (all of whom had partial involvement preoperatively) (Figure 6), and of the remaining 18 (60%) who had partial recovery, 9 (50%) developed aberrant regeneration (Figure 5).

FIGURE 6 Case 17. A 52-year-old patient was referred to the neuro-ophthalmology unit in June 1990 with a 3-year history of double vision and headache. He had undergone exploration of the middle cranial fossa in February 1987, but no biopsy was taken. MRI revealed (more ...)

Following cavernous sinus surgery, diplopia may not be an immediate concern because the eyelid is often ptotic due to either a third nerve palsy or swelling associated with the craniotomy. As the swelling resolves and lid function recovers, patients may complain of double vision. In the short term, the simplest way of dealing with this is by occlusion. At least initially, the deviation is usually incomitant enough not to respond to the use of prisms, although in highly motivated patients, an area of binocularity can be established or the eccentric area of binocularity can be moved to primary position. Fresnel stick-on prisms are the least expensive solution and permit treatment of larger deviations than ground-in prisms. Unfortunately, Fresnel prisms will blur vision. Botulinum toxin injections have been used in patients with cranial nerve dysfunction. There is no question that injections into the medial rectus in a patient with a complete sixth nerve palsy may delay development of medial rectus contracture and progressive esodeviation of the affected globe. It is unlikely, however, that botulinum toxin injections will result in functional binocularity in patients and probably should not be offered on that basis. There is also a reasonable incidence of recurrent ptosis associated with botulinum toxin injections due to leakage of the botulinum toxin as well as the potential for inducement of vertical movement problems on top of the already existing cranial nerve palsies.

From a functional point of view, the fourth nerve is certainly the easiest to deal with, and new involvement is probably the least concern. If this is isolated and stable, an ipsilateral inferior oblique weakening procedure or, alternatively, a contralateral inferior rectus weakening procedure should result in alignment.

The difference between complete and incomplete sixth nerve palsies is substantial from a functional point of view. Although it is possible to significantly increase the area of binocularity in patients with incomplete sixth nerve dysfunction with relatively simple, horizontal recession and resection procedures, those patients with complete sixth nerve dysfunction have a much poorer prognosis for complete resolution of diplopia. It is possible to align even these patients with transposition procedures, but the area of binocular single vision is usually significantly limited (Figure 5).

Although less commonly seen as an initial manifestation of cavernous sinus pathology, postoperative third nerve dysfunction may be the greatest challenge to ocular rehabilitation. Although one can argue that realignment of a patient with complete sixth nerve palsy is not trivial, this pales in comparison with an attempt to regain functional binocularity in a patient with a complete or even incomplete but severely compromised third nerve.

Surgical intervention for realignment should always be delayed until full recovery has occurred. Early cranial nerve dysfunction in our patients often cleared within a period of weeks to 1 or 2 months, but late improvement was also possible. Quantitative assessment of ocular motility with the use of Hess screen and binocular single vision field testing is probably the best means of assuring stability before consideration of muscle surgery. In patients with persistent cranial nerve dysfunction, often there is substantial residual limitation of globe duction. This is particularly true in patients who develop aberrant regeneration. Although it is possible to move the area of binocularity with muscle surgery on that eye, an increase in the binocular single vision requires procedures that limit rotation of the opposite eye. The Faden posterior fixation operation has been particularly useful in expanding areas of binocularity as well as moving them to more functional locations. By limiting the excursion of the normal eye, more extensive binocularity may be achieved (Figure 7).

FIGURE 7 Case 22. A 57-year-old right-handed patient was referred on January 29, 1991. At that time, she gave a history of 10 months of horizontal diplopia. When the double vision failed to clear, an MRI scan demonstrated a left-sided parasellar meningioma, and (more ...)

Recurrence can happen even with complete excision, and a 10% to 25% incidence has been reported.^{136,138,140,142} The rate of recurrence seems to increase with the duration of follow-up, and previous studies of meningiomas in general suggest that the longer the follow-up, the greater the recurrence rate. Mirimanoff and colleagues¹⁴³ reported recurrence rates for all meningiomas as 7%, 20%, and 32% at 5, 10, and 15 years of follow-up, respectively. Cavernous sinus numbers are likely to be higher.

The extremely slow growth and variable nature of these tumors are major factors in making it difficult to determine recurrence rates. Another cause for frequent recurrence is the invasion of the vascular wall^98,144,145 and the involvement of the cranial nerves themselves.¹⁴⁶ As seen in our cases, carotid and cranial nerve invasion occurs even with benign tumors and limits possible complete resection.

Progression rates are clearly higher if tumor is intentionally left along the carotid and around the cranial nerves. In reviewing a series of 225 subtotally resected meningiomas in all locations, Mirimanoff and colleagues¹⁴³ found progression in 37% at 5 years, 55% at 10 years, and 91% at 15 years following surgery. Mathiesen and associates¹⁴⁷ found that 42 (61%) of 69 patients (out of a series of 315 skull base meningiomas operated on between 1947 and 1982) who were treated with subtotal resection died of disease usually by 10 years. There is less data (and of shorter duration) on incompletely resected cavernous sinus meningiomas, but after 2 years of follow-up, O’Sullivan, and associates¹⁴⁰ reported 2 (6.5%) of 31 cases with progression, and De Jesus and colleagues¹⁴⁸ detected 7 (15%) of 46 patients who showed progression (mean follow-up, 34 months). The frequency of progression was up to 38% in those patients who had been followed for more than 5 years. Earlier studies also found that the extent of the original excision was the best prognostic factor determining recurrence.¹⁴⁹ These data support an argument for more aggressive attempts at complete excision, but the recurrence rates clearly indicate that even with total surgical resection, meningiomas may recur. Unfortunately, more aggressive surgery probably carries a higher risk of morbidity.¹⁵⁰

Recent advances in molecular genetics reveal chromosomal changes^151–153 that put patients at high risk for recurrence and progression. There is hope that these and other molecular genetic markers may indicate the risk of recurrence in the future.

In our series the acknowledged residual tumor and the high incidence of subsequent gamma knife treatment underscore the difficulty in achieving complete resection of lesions involving the cavernous sinus. It is often difficult, if not impossible, to make definitive statements about complete excision without more detailed long-term follow-up. The slow growth of many of these lesions continues to be a challenge to surgery in the parasellar region.

Unfortunately, recurrences are still possible even in completely resected nonmeningiomatous tumors. This is particularly true of invasive pituitary tumors in which complete surgical excision may not be possible. Eisenberg and colleagues¹⁵⁶ reported a 28% incidence of recurrence. Based on the results of transphenoidal surgery series for pituitary marcoadenomas, it is likely that longer-term follow-up will reveal even higher rates of recurrence.

Dolenc¹¹¹ noted one of 40 trigeminal neurilemomas recurring in 1994, and Taha and associates¹⁵⁷ found a 10% incidence of recurrence of trigeminal neurilemomas. Sekhar and coworkers¹⁵⁸ reported 4 (13.3%) of 30 nonmeningiomatous benign lesions recurring within the area of the cavernous sinus.

All of these numbers are substantially better than the statistics from series published before the development of skull base approaches. Recurrence rates of up to 65% have been reported in patients previously treated. This likely reflects not only the improvements in surgical approaches, but also the ability to better define the anatomy with the advent of neuroimaging studies. Although the numbers are an improvement, recurrence is still possible, even with reported complete clinical excision.

Unfortunately, the effect of the nature of the pathology on the incidence of neuro-ophthalmic complications has not been previously addressed. Somewhat surprisingly in this study, when the data was subject to rigorous quantitative assessment, there was little difference between the complication rates for patients that were found to have meningiomas and those with nonmeningiomatous pathology. This was particularly true for the chance of optic nerve damage. Nonmeningiomas did have a better chance of complete recovery of pre-existing cranial nerve palsies, although the incidence was low.

Unfortunately, but not surprisingly, no patient with complete optic neuropathy improved following surgical decompression in the area of the cavernous sinus.⁹⁸ This has also been demonstrated in the present study; none of the 4 patients with hand motion vision or worse, including 1 meningioma, experienced improvement. On the other hand, one could argue that decompression of the optic nerve might be accomplished without invading the cavernous sinus, inviting the attendant morbidity described.¹⁵⁹

Given that the majority of patients with cavernous sinus pathology present with evidence of ocular motor abnormalities, one of the hoped-for results of cavernous sinus surgery is an improvement in ocular motility. Early reports were encouraging. Lesoin and associates¹³⁷ reported improvement in ocular motor function in 10 (48%) of 21 patients. Sekhar and colleagues¹⁵⁸ reported improvement in 7 of 9 patients with third nerve palsies, 2 of 6 patients with fourth nerve palsies, and 6 (50%) of 12 patients with sixth nerve palsies in a series of mixed pathology. Somewhat more ominous was the report in 1991¹²⁰ that while 24 (44%) of 54 patients showed improvement in ocular motor function following cavernous sinus surgery, none of the 18 patients with meningiomas improved.

This poorer prognosis in meningiomas has been confirmed in subsequent reports. In a series restricted to meningiomas, DeMonte and associates¹³⁶ found improvement in ocular motor function in only 13%, whereas patients with nonmeningiomatous tumors reportedly had a 50% improvement.¹⁵⁶ In all patients, the third nerve was much less likely to recover than the sixth nerve.¹⁵⁶ The ability to improve preexisting ocular palsies was very limited in the present series. In only 1 patient was there felt to be improvement in a preexistent third nerve palsy (none cleared completely), and only 2 patients with sixth nerve palsies improved (clearing in one).

This classification system was used by Cusimano and colleagues¹⁴² to follow 124 patients who underwent cavernous sinus surgery. Four (27%) of the 15 who had poor preoperative binocularity improved to “good” or “excellent” status, and 6 (20%) of 30 patients with “good” preoperative binocularity improved to “excellent” status. The series included meningiomas and nonmeningiomatous tumors. Even in this series, there was little quantitative assessment.

The potential of completely excising a lesion and improving function must be balanced against the surgical potential for creating new problems as well as the natural history of the lesions themselves. The variability in the natural history of cavernous sinus tumors is a major limitation. In particular, menigiomas involving the cavernous sinus often have an extremely indolent history, exhibiting very slow progression. One of the difficulties in analyzing previous series that focus on the natural history is the variable inclusion criteria. The natural history is far worse if tumors arising from the clivus or petrous bone and secondarily involve the cavernous sinus are included, because these tumors will often compress the brainstem. If the series are restricted to tumors involving the cavernous sinus in isolation, the natural history is far more benign.

Lesser complications have been far more frequent and have included infarction with hemiparesis,^136,138,142 meningitis, and hydrocephalus. A CSF leak is one of the most frequently occurring complications following cavernous sinus surgery and is probably a direct result of surgical aggressiveness. Its incidence may be as high as 28%.¹⁴²

A sobering statement was made in Cusimano’s 1995 summary of the Pittsburgh¹⁴² experience: “the majority of patients in this series suffered a surgical complication.” Although he goes on to point out that most complications were minor or treatable, this potential must be weighed against the expected benefits of surgery. For psychological and other reasons, a case can still be made for aggressive surgery.¹⁶³

It is not the role of the ophthalmologist to discuss nonophthalmic risks with the patient without including the referring surgeon. Unfortunately, there is often a tendency for neurosurgeons to trivialize the morbidity in any transcranial procedure. As ophthalmologists are often the ones to recognize cavernous sinus pathology and thus refer patients to neurosurgery, it is useful to have a feel for the interaction of the neurosurgeon with the patient so that all parties are aware of the goals as well as the risks of any type of intervention.

As mentioned in our oculomotor results, none of the neurosurgical series discusses the concept of aberrant regeneration. The only time this is brought up is in those reports detailing intraoperative cranial nerve repair.^165–167 This is somewhat surprising given that meningiomas in the parasellar area may be responsible for the development of aberrant regeneration in the absence of a history of a third nerve palsy.¹⁶⁸ The combination of the classical findings of aberrant regeneration without a history of a third nerve palsy (primary aberrant regeneration) is almost pathognomonic of a meningioma or aneurysm in the parasellar region.

Because of the innervation of multiple muscles, damage to the third cranial nerve often results in incomplete resolution, with evidence of variable muscle misfiring. Not until the first third of the 20th century did explanations for this misdirection syndrome appear in the literature.¹⁷⁰ Morris Bender, a neurologist in New York, undertook experimental work with oculomotor nerve regeneration in which he sectioned the third nerve of a chimpanzee and followed its recovery.¹⁷¹ Walsh¹⁷² reviewed the importance of aberrant regeneration in 1947. It has been suggested that there may be alternative explanations for aberrant regeneration, including ephaptic transmission,¹⁷³ but it is likely that actual reinnervation of sprouting axons terminating in the wrong extraocular muscles is responsible for the majority of the findings. The classic finding of lid elevation with attempted depression and adduction is probably less consequential than the coinnervation of the superior and inferior rectus muscles, which leads to persistent limitation in the vertical gaze. Aberrant regeneration is probably underdiagnosed and is the responsible mechanism in the majority of patients with incomplete third nerve palsies and vertical compromise.

It is unlikely that the attempts at surgical repair of the third nerve during cavernous sinus surgery will reduce the incidence of aberrant regeneration. At this time, the best treatment for aberrant regeneration is to prevent its occurrence. Aberrant regeneration does not occur with presumed microvascular third nerve palsies, and its high incidence in patients undergoing cavernous sinus surgery suggests that perhaps there is more than simple vascular damage to the third nerve. It is possible that retained tumor within the nerve or direct surgical trauma results in misdirection in patients following cavernous sinus surgery. The possible development of aberrant regeneration has been one of the strongest arguments in favor of limiting surgery.^72,161

The current series has a much higher incidence of third nerve dysfunction. Only 1 (1.8%) of 57 patients showed no evidence of third nerve dysfunction immediately postoperatively. Exclusion of the 19 patients with preoperative third nerve dysfunction still results in a 97% incidence of acquired third nerve palsy immediately following cavernous sinus surgery. Although most palsies cleared rapidly, 4 (11%) of the 35 patients with long-term follow-up had complete third nerve dysfunction, and an additional 18 (51%) had an incomplete third. Nine (26%) demonstrated evidence of aberrant regeneration. Probably the closest numbers to the current series were reported by Cusimano and colleagues,¹⁴² who noted that 35 (60%) of 58 patients undergoing cavernous sinus surgery had binocularity that was “worse than excellent.”

Sindou and Pelissou,¹⁷⁴ in a literature review, reported on new fourth nerve palsies in 4 (7.3%) of 55 patients undergoing surgery for trigeminal neurilemomas. In Al-Mefty and Smith’s original report of 18 patients,¹⁶⁴ only 1 (5.6%) had a fourth nerve palsy. Nine (12.2%) of 74 patients in their expanded series⁹⁹ were said to have fourth nerve dysfunction (although the situation in 7 patients was listed as “too early to tell.”). Perneczky and colleagues¹²² found 2 (5.7%) fourth nerve palsies among 35 cases. The Pittsburgh group originally reported 9 (20.9%) temporary and 4 (9.3%) permanent fourth nerve palsies in 43 patients.¹⁵⁸ The incidence had changed to 7 (7.7%) of 91 cases¹⁶⁵ by 1991, and in an article later that same year, 11 (10.9%) of 101 cases.¹⁶⁹ In more recent series, new fourth nerve palsies occurred in only 1 (3.7%) of 27 patients undergoing cavernous sinus surgery.¹³⁶ Although the fourth nerve is theoretically at greater risk during surgery for meningiomas than nonmeningiomatous tumors within the cavernous sinus, Eisenburg and associates¹⁵⁶ reported 4 (11.1%) of 36 patients with new fourth nerve dysfunction.

In the current series, fourth nerve palsies could not be separated from other cranial nerve palsies. When the third and sixth nerve palsies were complete, the fourth nerve usually showed no function.

The sixth nerve is the most difficult to see during surgery because it is situated deep within the cavernous sinus, in relation to the carotid artery. Early cavernous sinus surgical series report worsening sixth nerve function in patients in whom it already existed. In the original series of cases from Pittsburgh, there were 12 preoperative abducens palsies and 14 temporary palsies postoperatively. In only 4 (9.5%) of the 42 cases was sixth nerve dysfunction said to be permanent.¹⁵⁸ When the series was updated to include 101 patients,¹⁶⁹ there were 10 permanent palsies (10%). Interestingly, a report earlier that year had revealed only 5 sixth nerve palsies (5.5%) among 91 patients undergoing surgery.¹⁶⁶ This difference raises the question of how closely the patients were monitored for evidence of sixth nerve dysfunction. Five (9.3%) of 54 cases in the series from Hanover had permanent sixth nerve dysfunction, and 10 had temporary abduction abnormalities.¹²⁰ Only one (1.4%) of 74 cases reported in 1991⁹⁹ was said to have a permanent sixth nerve palsy, although 7 were listed as “too early to tell.” Two (11%) of 18 patients with sixth nerve dysfunction preoperatively were reported to be worse following surgery.¹³⁶ When broken down into nonmeningiomatous benign tumors, there were still a number of new sixth nerve palsies following surgery (4 [14.3%] of 28 reported by Eisenberg¹⁵⁶).

There was no indication of the means used to assess cranial nerve function in any of these reports. None of these previously reported series include the numbers of patients with cranial nerve dysfunction seen in this current series. Only 8 patients (10%) had no evidence of sixth nerve dysfunction immediately after surgery. Although many of these dysfunctions resolved, 35 (69%) of 51 patients had some evidence of sixth nerve dysfunction at the follow-up more than 3 months after surgery, and 14 (27%) had a complete sixth nerve palsy.

Although there may be several explanations for the higher incidence of cranial nerve palsies in our postoperative patients, it is likely that quantitative detailed assessment may be largely responsible for explaining the higher numbers. Incomplete ophthalmoplegia is not incompatible with binocularity and good visual function, but may well lead to significant patient dissatisfaction. Future studies will need to include methods to evaluate functional outcome.

The potential for developing neurotrophic keratitis following the loss of fifth nerve function can be exacerbated by several factors. These include loss of tear production, seventh nerve dysfunction with problems in eyelid closure, and problems with ocular motility resulting in the inability to move the eye out of the palpebral fissure. None of these issues has been discussed in any prior neurosurgical series, however. Aggressive lubrication is an essential treatment in patients at risk for neurotrophic keratitis. The use of gold weight implants, tarsorrhaphy, or other protective procedures is probably best completed early when the loss of corneal sensation is associated with decreased seventh nerve function. A vascularization procedure (Gunderson flap) may be required to keep the eye from perforating with recurrent epithelial defects, erosion, or infection. Nerve growth factor treatment may be an additional mode of therapy in patients with persistent sensory loss.^176,177 Unfortunately, initial studies suggest that patients with surgically induced neurotrophic keratitis are probably the most resistant to treatment.

Over the last decade an increasing interest has been expressed for the use of radiosurgery. This is defined as a single-dose application, either of a gamma source or through a linear accelerator. Additional smaller studies of the use of proton beam have also been reported for meningiomas involving the cavernous sinus. The development of single-dose focal radiation therapy (LINAC^182,183) and gamma knife^184–194 has offered additional options. The exact role of these modalities is yet to be defined.¹⁹¹ Significant series reported on the use of gamma knife have included upward of 1,000 meningiomas involving the cavernous sinus treated either primarily or following subtotal resection.^{189,190,192–199} The emphasis has been on “control.”¹⁹⁷ Unfortunately, although some of the case follow-ups have been 8 years or more, the majority of series have been limited to 3 to 4 years.^{190,197,200,201} Based on previous experience, this is much too early in order to be able to truly assess the effectiveness of gamma knife in these patients.^{184–188,190}

While complications have been low, they do occur. Several studies have addressed the potential for radiation optic neuropathy in patients undergoing gamma knife. The optic nerve and the visual pathways are more susceptible to potential radiation damage.^202,203 Some investigators have emphasized the ability to treat lesions directly applied to the optic nerve.^201,204 On the other hand, it has been suggested that optic neuropathy could be as high as 77.8% when the optic nerve received a dose of 15 Gy or more.²⁰⁵ Therefore, lesions should be situated at least 3 mm away from the afferent visual pathways for safe treatment. The “safe dose” of radiation to the optic nerve is yet to be determined, and this fact argues for a conservative approach.²⁰⁶ The oculomotor and trigeminal nerves are relatively resistant to radiation therapy.^196,204,207 In spite of this, trigeminal and ocular motor problems have been reported.^208,209

In order to decrease the chance of damage, several investigators have suggested lowering doses.²⁰⁴ Others have emphasized the importance of treating the entire tumor to at least a dose of 14 Gy.²¹⁰ Because of the restriction of radiosurgery to a limited size, some investigators have continued to advocate fractionated radiation therapy for those tumors that are larger.¹⁹¹ Because of extremely low morbidity, several investigators still advocate the use of fractionated radiation therapy for cavernous sinus meningiomas in general. ^211,212 The use of gamma knife treatment with a particular interest for neuro-ophthalmologists was summarized by Carvounis and Katz in 2003.²¹³

During the last 2 decades several medical treatments have been proposed for meningiomas. Unfortunately, data on hormonal therapy²¹⁴ and hydroxyurea have been disappointing after initial optimistic reports. At this time there are no proven efficacious therapies for meningiomas beyond surgery and radiation. It is hoped that the recent mapping of chromosomal defects in meningiomas^151–153 will lead to alternative treatments.

In view of the slow progression of parasellar meningiomas, a number of surgeons, including several with significant clinical experience, have advocated a limited approach to these tumors consisting of resection of only the extracavernous portion possibly to be followed with radiation therapy.^{72,162,215,216} This offers the advantage of decompressing the optic nerve when involved, reducing the tumor size to something that can be treated with radiosurgery, and most important, limiting the complications of aggressive intracavernous surgery.^217–221 Their attitude might best be summed up by a statement made by Madjid Samii: “Why should we risk a carotid artery and eye function for a lesion that hardly ever endangers the patient’s life as long as the tumor part involving other structures such as the brain stem are resected.”

Pathology that affects the cavernous sinus is more common than previously thought, and recognition is largely due to the advent of modern imaging studies. Ophthalmic manifestations remain the most common indication of cavernous sinus pathology, although often patients are diagnosed when imaging studies are obtained for other reasons. The natural history of pathology affecting the cavernous sinus seems largely undetermined. Benign tumors may be asymptomatic for years, but progressive growth is common. Quantitative evaluation of the visual system is imperative to improve our understanding of the true natural history of these lesions. Failure to measure function accurately can lead to erroneous underestimation of the true incidence of dysfunction. Quantitative evaluation is important to direct therapeutic intervention. The timing of rehabilitative surgery depends on the degree of postoperative ocular motor stabilization.

Our study supports the literature that cavernous sinus surgery is technically feasible with relatively low mortality in patients with meningiomas. Nonetheless, quantitative assessment indicates that cavernous sinus surgery only infrequently benefits preexisting cranial nerve dysfunction. The one exception to this rule is extracavernous extension with compression of the optic nerve, which can be benefited by decompression in the area of the orbital apex. One caveat not previously reported in the literature is the danger to the optic nerve posed by the skull base pterional approach, which requires take-down of the anterior clinoid and portions of the optic canal. Although central acuity may remain fairly good, quantitative assessment of optic nerve function may detect a previously unrecognized postoperative incidence of visual field abnormalities. The mechanism of damage to the optic nerve is unknown, but may be related to the surgical removal of the anterior clinoid, during which there may be heat transfer to the optic nerve or, alternatively, hemorrhage-induced vasospasm. It is also possible that there may be direct mechanical damage to the optic nerve within the canal. Improvement in compromised optic nerve function could be obtained by simply excising the extracavernous portion of the tumor without the inherent risks of surgery within the cavernous sinus proper.

Oculomotor dysfunction following cavernous sinus surgery occurred in 97% of cases. This deficit is transient in the majority, but may also be permanent. The prognosis for complete resolution is better if the cranial nerve palsy is incomplete following surgery. Patients with complete oculomotor dysfunction postoperatively have an 84% chance of partially improved function with time, but a 32% incidence of developing aberrant regeneration, which is not necessarily incompatible with binocularity. Surgeons should discuss the potential for diplopia and the need for extraocular muscle surgery prior to proceeding with intervention.

Trigeminal dysfunction is common. The fact that neurotrophic keratitis can be expected in more than one-quarter of patients argues strongly for the importance of ophthalmic input into long-term follow-up of patients with cavernous sinus lesions. In this way, many of the consequences of neurotrophic keratitis may be prevented, as indicated in the section “Ophthalmic Rehabilitation.”

To keep this series uniform, only the results of a single surgeon were analyzed. Additional cavernous sinus surgeries performed by other neurosurgeons were excluded from this study. Although it is theoretically possible that other surgeons could have better results, this particular surgeon had more experience with cavernous sinus surgery than all others present and maintains the largest volume of cavernous sinus procedures. Given that many previously reported procedures were completed at other institutions, it is impossible quantitatively to assess ophthalmic function elsewhere. We can only conclude that more intense scrutiny from a neuro-ophthalmic point of view would reveal a higher incidence of ophthalmic pathology. Because neurosurgeons are not specifically trained to quantitatively assess ophthalmic function, it may be misleading to accept previous papers that report patient improvement based on subjective assessment. It is imperative that patients be presented with accurate data before making decisions with regard to this mode of therapy. With other treatment options potentially available, patients should be fully educated before a decision is made to perform this procedure.

This study would have benefited from a longer follow-up, but because many of the patients were coming from great distances, it was not possible to get them all back for a quantitative assessment. We chose not to accept nonquantitative data reported from other institutions because of the lack of detailed assessment, thus limiting the percentage of long-term follow-up. This is important if definitive statements are going to be made with regard to risk percentages. At this time, however, the thesis that cavernous sinus surgery has low ophthalmic morbidity is not justified. Patients undergoing cavernous sinus surgery can expect to have worsening of cranial nerve function that, even when transient, may leave them with substantial residual problems, including development of neurotrophic keratitis, and evidence of aberrant regeneration, which may produce persistent diplopia. There is a low incidence of improvement in cranial nerve function, and therefore patients should not be promised ocular motor improvement with less diplopia following surgery.

The last 2 decades of surgical refinement in dealing with skull base lesions has largely taught us what we can do. What we should do depends on a frank discussion of the options and a realistic assessment of the risks and benefits of surgery and its alternatives. It is likely that in the future, surgery will once again play a smaller, rather than a greater, role in treating cavernous sinus lesions. A combination of limited surgery and some form of radiation may reduce the morbidity associated with surgical treatment.¹⁹⁵ This course has been suggested and probably represents the current mainstream opinion in neurosurgery. At this time the lack of many other good options will still require the consideration of surgical options, and in the right case, surgery, especially if tailored to a realistic goal, may provide significant benefit to our patients.

Funding/Support: None

Financial Disclosures: None

For the last 12 years diagnostic files have been kept in a computerized database in the Neuro-ophthalmology Unit at the University of Virginia. A search of the database for patients with cavernous sinus involvement produced 347 records, including 230 women and 117 men with an overall average age at presentation of 51 years. Patients broken down by diagnosis revealed a relatively similar age range across pathology; both vascular and neoplastic lesions presented at a mean age between the late fifth and early sixth decades. When evaluated by subgroups, the carcinomatous lesions tended to appear later with a mean age of 58, while the neurilemomas were diagnosed earliest with a mean age of 31. As opposed to most of the early series published in the pre-imaging era, inflammatory lesions made up a tiny portion of this series. One of the potential problems in the diagnosis of inflammatory lesions prior to the advent of imaging studies was the presumption of inflammation when symptoms remitted with steroid therapy. A course of steroid therapy, in fact, was suggested as a way of diagnosing Tolosa-Hunt syndrome. Spector²²² pointed out the frequency of misdiagnosing inflammatory disease in the setting of neoplasia. In 1986 he reported 4 cases of neoplastic lesions originally diagnosed as having Tolosa-Hunt syndrome. A response to steroids was taken as evidence of inflammation. Unfortunately, it is quite clear that neoplastic processes, especially the lymphoproliferative disorders, but also including meningiomas and other benign tumors, may have some response to steroid therapy with resultant diminution of symptoms and signs. The distinction between inflammatory and neoplastic lesions may be quite difficult, exacerbated by possibility of variation²²³ or even spontaneous recovery of a neoplastic induced cranial nerve palsy.²²⁴ Even in the modern era of imaging, neoplasia may sometimes simulate an inflammatory etiology. Only 20 patients were inflammatory or infectious etiologies, including cavernous sinus thrombosis in 4 patients, mucormycosis in 2 patients, and aspergillomas in 3 patients. There were 7 patients with Tolosa-Hunt syndrome, and most of these were diagnosed earlier in the series.

Aneurysm and fistulae continued to make up a large number of the patients within this series, representing 13% and 15% of the total, respectively.

The largest group pathologically was neoplastic. This is dramatically different from the series published between 1922 and 1977.^81,83,86,225 In Rucker’s 1958 series,²²⁶ of 1,000 cases of ocular motor paralysis, neoplasms made up only 17% of the cases. Of those series that had a significant number of mass lesions, malignancy was the most common, especially nasopharyngeal carcinoma with secondary involvement of the skull base. In the most recent large series of cavernous sinus lesions, published by Keane in 1996,⁸⁷ tumors made up the largest single group (30%), but the majority were still malignant, including nasopharyngeal carcinoma and metastatic disease. Only 4 cases were meningiomas (9% of the 45 tumors). This was similar to 3 (6%) of 50 cases in Jefferson’s series⁸³ and 3 (4.3%) of 70 cases in the Mayo Clinic series published by Thomas.⁸⁶ As late as 1982, Parkinson¹³³ still felt that malignancy was the most common tumor involving the skull base.

Malignant tumors still make up a significant percentage of cavernous sinus lesions in our series (48 [13.8%] of 347). Some of these were due to direct extension from the nasopharynx and the paranasal sinuses; more were metastatic or due to neurotrophic spread. The preponderance of meningiomas in our series reflects a referral bias (in view of the skull base interest at the University of Virginia, Charlottesville) and the true frequency of meningiomas. These numbers are similar to other large surgical series, including those from Pittsburgh,²²⁷ University of Arkansas,⁹⁹ and Ljubljana.⁹⁸ The other likely reason for the change in frequency relates to the advent of modern imaging studies. Meningiomas affecting the cavernous sinus may be quite indolent with little in the way of symptoms. As the use of magnetic resonance imaging (MRI) and computed tomography (CT) for evaluation of various cranial complaints increases, the number of meningiomas fortuitously discovered will likely increase. Many of the patients in our series were diagnosed when a scan was done to evaluate headaches. Headache and facial pain was the second most frequent symptom in the series of 82 patients evaluated for cavernous sinus surgery (see above).

The other difference between our series and that of Keane was the relative decreased frequency of trauma as a specific etiology. Although several of our carotid cavernous fistulae were traumatic in etiology, trauma as a primary cause of the cavernous sinus syndrome was diagnosed in only one case. This contrasts to Keane’s finding of trauma as an etiology in 36 patients, or 24% of his series. His series was drawn from patients presenting to a Level 1 trauma facility. In addition, 17 of Keane’s patients had surgical trauma following craniotomy and parasellar surgery.

Another important distinction between this series and Keane’s was his requirement that patients demonstrate multiple cranial nerve palsies. In this series patients were included on the basis of radiographic as well as clinical characteristics. Thus, several of the patients in the University of Virginia series had no evidence of ophthalmic or neuro-ophthalmic findings, and only imaging studies indicated involvement. In the earlier portion of our series, many of these patients were diagnosed on CT scan alone, but more recently, patients with cavernous sinus pathology were best evaluated by MRI with the use of gadolinium contrast.

Neoplasia	225
Malignant	48
Carcinoma	25
Bone-based tumors	14
Sarcoma	2
Other	7
Benign	177
Meningioma	118
Pituitary	35
Neurilemoma	6
Cavernous hemangioma	6
Other	6
Vascular	97
Aneurysm	45
Fistulae	52
Inflammatory	20
Tolosa-Hunt	7
Thrombosis	4
Aspergilloma	3
Zoster	2
Mucormycosis	2
Sarcoid	1
Granuloma	1
Other	5
Total	347