Risk for Hemorrhage

The risk for hemorrhage from an incidentally discovered CM is controversial. Naturally, this risk is a function of how hemorrhage is defined and whether these lesions are assumed to be present at birth. We define a hemorrhagic event as a clinical history of an apoplectic episode or evidence of subacute or acute blood products on computed tomography or MRI. The characteristic hypointense ring seen on T2-weighted MRI, however, is due to hemosiderin and, in our opinion, does not define an acute hemorrhagic episode.

Kondziolka and coauthors reported prospective hemorrhage and rehemorrhage rates of 2.4% to 5% per year, respectively.⁹ In contrast, in our institutional retrospective review, hemorrhage and rehemorrhage rates were 5% and 30%, respectively.¹⁰ Regardless, the timing of a subsequent hemorrhage is impossible to predict, with the interval between hemorrhages ranging from hours to years.

Several factors have been proposed to predispose a CM to rupture, including its location,^7,11 a history of previous rupture, its size,^12,13 and the presence of an associated developmental venous anomaly.¹⁴ The factor most consistently associated with increased risk for rupture across series is location. The hemorrhage rate of infratentorial lesions may be 30 times that of lesions in the supratentorial compartment. Both retrospective and prospective studies undertaken to define risk factors for hemorrhage from CMs have consistently identified the location of a lesion as having a significant impact on the rate of rupture. Brainstem CMs consistently have a higher rate of symptomatic hemorrhage than those at other locations. Hemorrhage rates as high as 60% have been reported for brainstem CMs.¹⁵

The mechanism for such a disparity in rupture rates, however, remains obscure. Most authors attribute this difference, at least partially, to the sensitivity of the brainstem to hemorrhage. In the literature, a history of previous rupture is strongly associated with as much as a sevenfold increase in the risk for prospective rupture.^9,16

Some authors have attempted to link the presence or absence of an associated venous malformation with a higher rate of rupture.¹⁴ In our experience, though, CMs have universally been associated with venous anomalies, whether supratentorial, infratentorial, or even extra-axial (e.g., for CMs of the cranial nerves). Venous malformations are completely benign, but abnormal constellations of veins that drain normal brain tissue. They are the most frequent form of vascular malformation and are a common incidental finding on MRI.

It is important to emphasize that venous malformations, per se, do not rupture; however, they are frequently associated with CMs that do.¹⁷ Thus, given the association between CMs and venous anomalies, it is currently thought that any hemorrhage in the vicinity of a venous anomaly is the result of rupture of an associated CM, regardless of whether it is visualized on imaging studies (some CMs may be small enough to be missed on routine imaging studies). Unfortunately, a complete consensus on this point is lacking. Because of the association between CMs and venous anomalies, the older literature is replete with suggestions that venous anomalies may occasionally hemorrhage.¹⁸

Observation

Conservative management consisting of observation and repeated neuroimaging is appropriate for patients whose symptoms resolve completely after an acute hemorrhagic event or for patients with incidentally discovered lesions. Patients often express concern regarding activities or medications that may exacerbate the hemorrhagic tendency of CMs. Patients harboring incidental CMs should be advised that the risk for hemorrhage from an incidental lesion is extremely low (some studies suggest less than 1%) and that the chance of a seizure disorder developing (with supratentorial lesions) may be 2% to 3% per year. Furthermore, besides pregnancy, no other factors are known to be likely to increase the risk for hemorrhage. Patients should not restrict their activities or exercise based on a theoretically increased risk. To date, no evidence suggests that there is a protective effect from avoiding even strenuous exercise. Anticoagulation is not contraindicated in these patients, although the risk of having a slightly more symptomatic hemorrhage may arguably be increased relative to normal.

Patients should also be reassured that for supratentorial lesions in ineloquent locations, hemorrhage from CMs is exceedingly unlikely to be catastrophic and rarely, if ever, causes death. A benign manifestation consisting of headache is much more likely. This statement is not necessarily true, however, for deep-seated lesions, including infratentorial CMs or those located supratentorially in eloquent cortex (e.g., motor strip).

Radiation Therapy

Radiation therapy, including stereotactic irradiation, has not been shown to confer a protective benefit from hemorrhage in the treatment of CMs, partially because the natural history of CMs has remained difficult to define and because CMs treated with radiosurgery do not disappear on subsequent imaging studies. Furthermore, the hemorrhage rate never approaches zero. In fact, radiation has been implicated in the pathogenesis of these lesions. Moreover, for CMs in eloquent regions such as the brainstem and basal ganglia, new neurological deficits related to a treatment effect have been reported in 17% to 59% of patients.^19,20 At present, therefore, most authors do not recommend radiation therapy for the treatment of deep-seated CMs.

Surgical Indications

The indications for surgery depend on the lesion’s location and symptoms. For lesions causing medically refractory epilepsy, surgery may be indicated to reduce or eliminate the seizures when a seizure focus can be reliably determined. Although most supratentorial CMs that do not cause seizures can be observed safely, there are several important exceptions. CMs located anywhere in the ventricular system should be considered for resection. A ventricle would not be expected to provide sufficient tissue pressure to tamponade bleeding from a ruptured CM. Hence, patients would be at risk for a devastating intraventricular hemorrhage. Furthermore, the behavior and surgical management of deep-seated supratentorial CMs, such as CMs of the thalamus or basal ganglia, are more akin to those of brainstem CMs than to other supratentorial lesions. They are consequently managed with the same considerations in mind.

For these deep-seated lesions, including brainstem CMs, surgical resection is appropriate if the CM reaches a pial surface, hemorrhages repeatedly in association with progressive neurological deficits, is manifested as acute hemorrhage outside the lesion capsule, or causes a significant mass effect from large intralesional hemorrhages. We recommend avoiding a myelotomy through even the thinnest amount of tissue in the floor of the fourth ventricle. Typically, we resect only clearly exophytic lesions in this location. The thickness of the rim is best ascertained on T1-weighted MRI. Lesions that clearly reach a pial surface on T1-weighted imaging can be considered for resection.

Surgery is indicated for posterior fossa locations outside the brainstem (i.e., cerebellar hemispheric CMs) in the event of acute hemorrhage producing a mass effect, for CMs that have ruptured multiple times, and for lesions exerting a significant mass effect from intralesional hemorrhage or expansion of the lesion. It is unnecessary for cerebellar CMs to approach a pial surface for it to be resected without causing morbidity. Similarly, lesions arising from the middle cerebellar peduncle that are exophytic into the fourth ventricle may be resected safely (we prefer the telovelar approach for these lesions, see later).

To facilitate the removal of acute hemorrhage, we typically wait 3 to 5 days for the hematoma to liquefy. If the patient is deteriorating rapidly, however, the brainstem may need to be decompressed in an emergency fashion. Acute hematomas tend to be tenacious and to require more manipulation of the surrounding parenchyma than do more subacute, yet liquefied, clots.

Contraindications

Patients should not be considered for surgical therapy if they have severe concomitant medical problems or if they have had a single hemorrhagic episode from a brainstem CM in an unfavorable location (e.g., far from a pial surface). In our experience, these patients can be managed conservatively until they suffer at least one more hemorrhage, especially if even the thinnest rim of tissue in the floor of the fourth ventricle must be traversed. If fixed deficits develop after another hemorrhage, surgery may be offered as a treatment option even for intrinsic brainstem lesions.

For patients with multiple CMs who become symptomatic with seizures, surgery should not be considered unless an individual seizure focus can be identified. Furthermore, although rare, the presence of multiple brainstem CMs is a relative contraindication to surgical intervention unless the patient’s initial signs and symptoms can be confidently attributed to a single lesion.

Finally, for posterior fossa lesions in general and for brainstem lesions in particular, surgeons should not hesitate to abort surgery if unexpected bleeding is encountered, significant changes in neuromonitoring develop, or unexpected intraoperative events occur. Even under ideal operative conditions during the treatment of these lesions, the potential for significant morbidity is high. Therefore, surgeons should have a low threshold for aborting procedures and returning at another operative sitting if necessary.

Goals of Surgery and Patient Counseling

For posterior fossa and other deep-seated lesions, the goals of surgery are to extirpate the CM completely while minimizing the amount of normal eloquent (i.e., brainstem or thalamus) tissue traversed. The associated venous anomaly or malformation should be preserved. If a large venous malformation is occluded, venous infarction may result. Superficial supratentorial CMs can usually be resected completely with minimal morbidity and excellent outcomes. Nevertheless, in these cases, too, every attempt should be made to preserve the associated abnormal venous drainage. Under image guidance, a tailored craniotomy is generally sufficient to expose these lesions. No attempt to resect the surrounding hemosiderin-laden brain is undertaken, regardless of whether the CM is in the supratentorial or infratentorial compartment.

Once the decision to operate has been made, appropriate preoperative counseling is critical, especially for patients with deep-seated lesions. Patients should be educated that their deficits are likely to worsen after surgery but will typically improve with time. Patients should be told that the surgical experience is similar to having another hemorrhage. They should be warned, if appropriate, that a tracheostomy or feeding tube may be necessary on a short-term basis and that a moderate course of rehabilitation will probably be necessary. Such information eases patients’ anxiety and provides them realistic expectations about the process.

To determine the best surgical approach, we use the “two-point method” (Fig. 394-2).²¹ One point is placed in the center of the lesion and a second is placed where the lesion most closely reaches a pial surface. The two points are connected, and the resultant straight line through the least eloquent tissue dictates the most appropriate surgical approach. Preoperative permanent neurological deficits, such as seventh or eighth cranial nerve palsies, can also influence the choice of approach. Such deficits, for example, may make a translabyrinthine or transcochlear approach more attractive.

FIGURE 394-2 Artist’s rendition of the “two-point” method applied to medullary cavernous malformations in similar locations. A first point is placed in the center of the lesion and a second at the point where the lesion comes closest to the surface. A line connecting the two points determines the optimal approach. The two lesions require different surgical approaches, the suboccipital (left) and the far lateral (right), because the points where they reach the surface are different.

(Used with permission from Barrow Neurological Institute.)

We avoid transcortical approaches whenever possible. Importantly, when using the two-point method to access deep-seated supratentorial lesions (e.g., thalamic CMs), an ependymal surface can be substituted for a pial surface. Hence, the two-point method might be used to select a posterior interhemispheric transcallosal approach for such a lesion.

Preoperative Imaging

As mentioned, the MRI appearance of CMs is pathognomonic and hence sufficient for surgical planning. Angiography is not indicated. Preoperative gadolinium-enhanced, contiguous, nonoverlapping, 1-mm axial-slice, single-echo MRI is routinely obtained for frameless stereotactic guidance. We use Stealth (Medtronic, Minneapolis, MN) for intraoperative image guidance integrated into the operative microscope such that the surgeon’s focal point with respect to the lesion’s location and operative trajectory is displayed.

Intraoperative Monitoring

Intraoperative monitoring is a valuable adjunct to help minimize complications during craniotomy. At our institution, monitoring of somatosensory evoked potentials (SSEPs) and compressed spectral analysis or electroencephalography is routinely performed. For brainstem lesions, motor evoked potentials and brainstem auditory evoked potentials (BAERs) are also monitored. These monitoring techniques should be applied before and after the patient is positioned because excessive flexion or rotation of the neck can cause disastrous outcomes such as vascular compression, brachial plexopathy, spinal cord injury in the presence of spondylosis, or excessive venous pressure.

Although these techniques provide continual feedback, postoperative neurological deficits are not always preceded by a change in the recorded waveforms. Both false negatives and false positives can occur. Consequently, the pathway being monitored must be relevant to the operation. Baseline recordings are useful so that changes can be evaluated as relative rather than as absolute. Surgeons can then make adjustments accordingly. BAERs are central signals in that they relate to the cochlear nucleus and can be monitored during surgery on an intrinsic pontine lesion. Intraoperative changes in wave latencies suggest an interruption in the auditory pathways. However, damage to motor or other cranial nerve nuclei can go undetected. When the floor of the fourth ventricle is involved, the facial colliculus can be stimulated to localize it accurately and to minimize the chance of compromising the function of the seventh or eighth cranial nerves. Disciplined and meticulous surgical technique is the best means for avoiding postoperative deficits. We have found no benefit in performing more extensive or invasive physiologic techniques such as motor mapping.

Surgical Technique

In general and regardless of location, CMs are accessed through minimal cortical openings. The CM is removed sharply and in piecemeal fashion. If intrinsic lesions fail to reach a pial surface of the brainstem, normal brainstem tissue will be violated during surgery. In this case, an opening is made by using hemosiderin staining or a bulge in the brainstem as a guide. Alternatively, the two-point method may be applied in conjunction with frameless stereotactic guidance. Entry into the brainstem is well tolerated, even in the case of deep-seated intrinsic lesions, if the cortical opening is small and the fibers of the brainstem are gently stretched to allow resection of the lesion. In contrast, exophytic lesions are readily apparent, assuming that the correct surgical approach was chosen. Lesions usually have a characteristic “mulberry” appearance with a thin layer of arachnoid (Fig. 394-3).

FIGURE 394-3 An exophytic cavernoma of the medulla with the classic “mulberry” appearance and a concomitant associated venous malformation to the right of the lesion.

(From Spetzler RF, Koos WT. Color Atlas of Microneurosurgery. Vol. 3. Intra- and Extracranial Revascularization and Intraspinal Pathology, 2nd ed. New York: Thieme; 2000:370.)

The CM can be entered with bipolar cauterization. Acute, subacute, and chronic blood products can be suctioned. The capillary network or hemangiomatous portion can then be gently dissected with microdissectors while the surrounding parenchyma is preserved. Microscissors may be necessary to detach the lesion from surrounding tissue. During dissection the surgeon should be mindful of the ubiquitous venous anomaly. If a large, associated venous malformation is entered and coagulated, venous infarction and a devastating outcome can result. Smaller venous tributaries, however, can generally be coagulated and transected with impunity. If a surgeon is unsure of their significance, associated veins within the resection cavity should be preserved.

Role of Intraoperative Magnetic Resonance Imaging

It may be difficult to determine the completeness of resection of a CM, especially a deep-seated lesion, when the cortical opening is very small, illumination is relatively poor, and the surgical resection cavity may be incompletely visualized. Again, frameless stereotaxy with a state-of-the-art microscope is essential to maximize the chance of complete removal. Despite using contemporary techniques, however, the surgeon cannot always be certain that the lesion has been resected completely without damaging surrounding structures.

Recently, we have used intraoperative MRI with a 3-T magnet in patients undergoing surgery for CMs. Our experience thus far suggests that this tool will prove useful in the surgical management of this disease, particularly brainstem CMs. More experience is necessary, however, before conclusive recommendations about the optimal role of intraoperative MRI can be made.

Postoperative Management

For superficial supratentorial lesions, the postoperative management of patients undergoing surgery for resection of CMs is similar to that for patients undergoing surgery for tumors in similar locations. Patients who have a brainstem CM resected often remain intubated the first 24 hours after surgery. After they demonstrate a good cough and gag reflex in the intensive care unit, they are extubated. Evaluation of postoperative swallowing is recommended for patients whose function could be at risk because of the lesion’s location. When these functions appear less than optimal, short-term (months) tracheostomy or feeding tubes should be placed. Such need should not be interpreted as treatment failure because few patients require these adjuncts on a long-term basis. If patients are warned of the potential need preoperatively, they tend to accept these procedures more readily after surgery.

If the patient is stable, we perform MRI on postoperative day 1 to assess for the presence of residual lesion and to serve as a baseline for comparison with future studies. Hemosiderin staining makes assessing the extent of resection difficult. Consequently, follow-up imaging should be performed annually for the first few years to monitor for progression or recurrence.

Despite postoperative MRI findings consistent with cure, recurrence rates can be as high as 5%. Thus, the physician should always be suspicious for residual lesions. However, venous malformations associated with hemosiderin staining can also mimic CMs. Furthermore, in patients who undergo “complete resection,” recurrence can develop years later. This trend may reflect the presence of an unrecognized residual lesion at surgery or a de novo lesion that developed as a result of an associated venous anomaly. Careful annual clinical and imaging follow-up should be instituted for all patients. Symptomatic patients may need to be evaluated even sooner. If patients are free of symptoms and their MRI study is negative, follow-up intervals can be increased at the surgeon’s discretion.

Midline Suboccipital and Telovelar Approaches

The midline suboccipital approach (Fig. 394-4) is used to reach lesions situated in the cerebellum, the posterior cervicomedullary junction, and the midline floor of the fourth ventricle. In this approach, the patient is placed prone on chest rolls or a laminectomy frame, and the neck is flexed to open the space between the foramen magnum and C1. After a strip shave, a midline skin incision is made that extends from approximately C3 to the inion. The fascia is opened such that a Y-shaped cuff with its base attached to the inion is created. This technique helps identify the avascular midline plane between the semispinalis capitis, trapezius, and splenius capitis muscles and provides a watertight fascial closure at the end of the procedure. The posterior cervical musculature is elevated from the suboccipital bone by subperiosteal dissection and retracted laterally with fishhooks. A suboccipital craniotomy is fashioned with a pneumatic drill that has a side-cutting bit and footplate. A burr hole can be made over the cerebellar hemisphere. Alternatively, the footplate can be placed under the foramen magnum. The dura is opened in a Y-shaped fashion, with its base along the torcular. The telovelar variation of the suboccipital approach is ideal for lateral fourth ventricular lesions, especially those involving the middle cerebellar peduncle. In this approach, the patient is similarly positioned prone on the operating table. The surgeon sits at the head of the operating table. Maximal flexion, particularly at the occipitocervical articulation (i.e., capital flexion), is even more important than with a routine suboccipital approach because the angle of approach to the lesion is much more obtuse. For this reason also, the head of the bed may need to be lowered considerably. After performing the same soft tissue dissection as for a conventional suboccipital approach, the posterior arch of C1 is removed piecemeal with a rongeur or with a side-cutting bit and footplate attachments on a pneumatic drill. The suboccipital craniotomy and dural opening are fashioned as described earlier. A retractor is useful to elevate the cerebellar vermis or hemisphere (or both) during dissection of the cerebellomedullary fissure, which is done sharply, until the cerebellar peduncle is visualized.

FIGURE 394-4 A, Extent of the craniotomy (dashed line) required for the suboccipital approach. Fishhooks rather than traditional retractors are used for muscle retraction to bring the surgeon’s hands closer to the operative field. B, The dura is opened in a Y-shaped configuration (green dashed line). The inferior limb of the incision is just off midline to avoid the occipital sinus and to ensure that it is crossed only once.

(Used with permission from Barrow Neurological Institute.)

Finally, although lesions of the posterior midbrain can also be accessed by splitting the superior cerebellar vermis or superior medullary velum through a midline suboccipital approach, they are better addressed with the supracerebellar infratentorial approach (see later).

If a lesion is to be approached through the floor of the fourth ventricle, the location of cranial nerve nuclei becomes relevant. In 1993, Kyoshima and colleagues described “safe entry zones” above and below the facial colliculus.²² Bogucki and associates later modified the zones to access intrapontine lesions.²³ The infrafacial zone has a line 2 mm lateral to the median sulcus on its medial border, the hypoglossal triangle on its inferior border, the facial colliculus on its superior border, and the vestibular area laterally. The suprafacial zone has the following boundaries: laterally, the superior cerebellar peduncle; medially, a vertical line 2 mm lateral to the median sulcus; superiorly, the frenulum veli; and inferiorly, the facial colliculus. Tumors, however, can distort normal anatomy and obscure these landmarks. In such cases, the facial colliculus can be stimulated directly while the facial nerve is monitored. Typically, however, lesions of the floor of the fourth ventricle are not treated surgically unless the patient has a complete neurological deficit such as internuclear ophthalmoplegia, abducens palsy, or facial paresis.

At the end of the procedure, the dura and fascia are closed in layers and the bone flap is replaced. Fibrin glue can be used on the dural suture line and bone cement in the craniotomy defect to minimize the chance of leakage of cerebrospinal fluid (CSF). During fascial and muscle closure, it is useful to release the head holder. Otherwise, significant spasms can occur and preclude adequate reapproximation of the suboccipital musculature to the occiput.

Orbitozygomatic Approach

The orbitozygomatic approach (Fig. 394-5) is used to access the anterior and lateral midbrain, interpeduncular region, rostral pons, pontomesencephalic junction, optic chiasm, hypothalamus, and the area caudal to the midthird ventricle. In 1986, Hakuba and coworkers first described this approach for lesions of the parasellar region, interpeduncular fossa, medial sphenoid wing, Meckel’s cave, and basilar tip.²⁴ The main advantage is that it permits downward retraction of the globe of the eye to gain an upward and oblique view of the interpeduncular fossa and third ventricle. The technique, as performed at our institution, has been detailed elsewhere²⁵ and is reviewed here briefly.

FIGURE 394-5 A, Lateral view of the osteotomies required for an orbitozygomatic approach. An anterior subtemporal notch can be made to facilitate a connection between the fifth and sixth cuts. B, An intraorbital perspective shows the osteotomies required. C, Lateral view of the exposure after the sequence of osteotomies (as numbered) has been completed and the bone flap removed (cross-hatched areas).

(Used with permission from Barrow Neurological Institute.)

The patient is positioned supine with the head rotated 30 to 60 degrees. Less rotation provides access to lesions of the posterior fossa. Slight extension of the neck, so that the malar eminence is the most superior point in the operative field, causes the frontal lobe to fall away from the anterior cranial fossa. The scalp incision extends from the root of the zygoma, 1 cm anterior to the tragus, to the midline or medial contralateral line. Alternatively, a bicoronal skin incision can be used. A vascularized pericranial flap should be preserved during the opening for use at the end of the procedure to isolate the exenterated frontal sinus should it be violated during the osteotomy. The temporalis fascia is incised along the posterior border of the skin incision and extended anteriorly just below the superior temporal line.

A dissection plane is then developed between the temporalis fascia and muscle, below the second fat pad. Next, the frontozygomatic and temporal zygomatic processes and superior orbital rim are exposed by elevating the temporalis fascia off the respective bony surfaces with a periosteal elevator. The periorbita is freed from within the orbit with a curet, and thin Telfa (Johnson & Johnson, Raynham, MA) strips are left in place to protect the periorbita. The elevator should not be passed too deeply toward the cone of the orbit because the optic nerve can be damaged. After the temporalis muscle is elevated and retracted from the squamous temporal bone, a standard pterional craniotomy is performed, followed by the orbitozygomatic osteotomy. Peripheral tack-up holes are drilled around the craniotomy and sutures are placed. The temporalis muscle is again released and placed over the craniectomy defect in preparation for the orbitozygomatic osteotomy.

For the conventional orbitozygomatic approach, an oscillating saw is used to make six osteotomies. The first osteotomy is made at the base of the zygoma by placing the saw on the zygomatic root just above the temporalis muscle. The blade is oriented anteriorly, medially, and obliquely. It must not be placed too deep to avoid violating the capsule of the temporomandibular joint. The second cut begins just anterior and inferior to the temporal process of the zygomatic bone and proceeds laterally to medially while directed slightly inferiorly. It stops in the midportion of the malar eminence at the zygomaticofacial foramen, which is connected with the third cut. This latter cut extends from the inferior orbital fissure from within the orbit through the orbital surface of the temporal bone and connects the second cut at an apex. First, the inferior orbital fissure can be identified with a No. 4 Penfield dissector, after which the tip of the saw can be placed. An inverted V is thus created on the malar eminence; the left and right links are the second and third cuts, respectively. The fourth cut extends through the orbital surface of the frontal bone posteriorly toward the superior orbital fissure. If more medial access is required, the supraorbital nerve can be mobilized from its foramen with an osteotome. If the foramen is high above the orbital rim, the nerve can be sacrificed, but forehead numbness will result. The fifth cut is made from outside the lateral orbital wall. It extends posteriorly from the inferior orbital fissure across the greater wing of the sphenoid bone and through the posterior orbit. The final (sixth) cut extends from the fifth cut to the superior orbital fissure. Alternatively, it can connect the fourth and fifth cuts just proximal to the superior orbital fissure. A rongeur can be used to remove the residual bony island of the greater wing of the sphenoid until the dural fold of the superior orbital fissure is identified. Importantly, the orbitozygomatic approach can be tailored to suit individual circumstances (modified orbitozygomatic approach) when a lesion requires less rostral visualization. However, for most deep-seated lesions, complete removal of the orbital roof and zygoma is indicated.

To ensure precise reapproximation at closure, holes can be drilled for a cranial fixation plate before the osteotomy is removed. Typically, 1.5-mm or smaller cranial fixation plates are used.

The dural opening extends from the medial superior orbital margin to the temporal tip in a semilunar fashion. It is tacked anteroinferiorly, which permits downward retraction of the globe. Tacking sutures are placed deep toward the orbital apex and anchored around the secure fishhooks. The operating microscope is then brought into the field, and a transsylvian or subtemporal approach is undertaken in atraumatic fashion. The uncus can be retracted posteriorly to allow access to Liliequist’s membrane and the interpeduncular fossa. The working distance to lesions in the parasellar region and the interpeduncular fossa is about 3 cm shorter with an orbitozygomatic approach than with a standard frontotemporal approach. A more upward and oblique view of the sylvian fissure, third ventricle, and upper brainstem can be achieved with less retraction on the temporal and frontal lobes. Patients should be forewarned of significant postoperative periorbital edema and diplopia, which usually resolve within a week.

On postoperative day 1, patients can begin jaw exercises to avoid restricted range of motion of the temporomandibular joint as a result of scarring of the temporomandibular ligament and joint capsule. Complications can include temporary or permanent paresis of the frontalis muscle, temporary orbital swelling, enophthalmos of an entrapped globe, cranial neuropathy, and leakage of CSF. Cosmetic results are usually good or excellent. Atrophy of the temporalis muscle can be minimized by using monopolar cauterization judiciously and by reapproximating the superior aspect of the temporalis muscle to a small muscle cuff of fascia left along the superior temporal line.

Retrosigmoid Approach

The retrosigmoid approach (Fig. 394-6) offers access to the posterolateral pons, lateral middle cerebellar peduncle, superior lateral medulla, and cerebellopontine angle. The patient can be positioned lateral, prone, or supine with a sandbag beneath the ipsilateral shoulder. In the supine position, the head is turned flat, parallel with the floor, and the neck is flexed such that a finger can be placed between the mandible and clavicle. Excessive rotation or flexion during positioning can cause neurovascular compromise, especially in patients with extracranial carotid disease or degenerative cervical spondylosis. Baseline SSEPs should therefore be recorded before and after positioning. If the SSEPs change, the patient should be repositioned. Alternatively, the patient can be placed prone or in the lateral decubitus position.

FIGURE 394-6 In the retrosigmoid approach, a burr hole is made at the transverse-sigmoid junction, and the craniotomy (dashed line) is made posterior and inferior to the burr hole.

(Used with permission from Barrow Neurological Institute.)

The skin incision starts above the auricle and curves behind the ear inferiorly 4 to 6 cm behind the mastoid and 6 to 8 cm behind the external auditory canal. Inferiorly, it extends just below the mastoid tip into the sternocleidomastoid muscle. Subperiosteal elevation of the muscle should reveal the digastric groove. If the incision is placed too far anteriorly, the scalp, muscle, and bone obscure visualization of the cerebellopontine angle.

The sinuses can be identified with frameless stereotactic guidance. Surgical judgment should always prevail inasmuch as the surgeon should be familiar with surface skull landmarks that approximate the locations of the major sinuses. The asterion is an unreliable external landmark for the transverse-sigmoid junction, but a line drawn from the root of the zygoma to the inion (the superior nuchal line) is a good approximation of the transverse sinus. Typically, the transverse-sigmoid junction is avoided by placing the burr hole below this line, 2 cm below the asterion, two thirds of it behind and a third of it in front of the occipitomastoid suture. After the burr hole is completed, a craniotomy can be fashioned behind the sigmoid sinus and below the transverse sinuses. The bone up to the transverse and sigmoid sinuses can be rongeured to expose their edges. Alternatively, the mastoid air cells can be drilled directly to expose the sigmoid and transverse sinuses and then the craniotomy can be performed. A line connecting the junction of the squamosal and parietomastoid sutures to the tip of the mastoid process approximates the course of the descending portion of the sigmoid sinus. The bone over the sigmoid and transverse sinuses can be drilled, first with a cutting burr and then with a diamond burr, and a thin shell of inner cortical bone left. The remaining thin shell of bone can be removed safely with a Penfield dissector or curet. If the sinus is entered during drilling, Gelfoam (Upjohn, Kalamazoo, MI) or Nu-Knit gauze (Johnson & Johnson, Arlington, TX) should be laid on top of the hole, followed by a cottonoid, but not inserted into it. Mastoid air cells violated during the exposure should be obliterated with bone wax to avoid leakage of CSF. To relax the cerebellum, some surgeons prefer to place a lumbar drain before surgery to release CSF while the dura is opened. Alternatively, a small linear incision, angled in the direction of the jugular bulbar, can be made over the cerebellar hemisphere. Strips of Telfa sponge (Kendall, Mansfield, MA) are placed on top of the hemisphere, and CSF is released after the arachnoid over the cerebellomedullary cistern is opened. Once the dura is opened, CSF must be released quickly because the cerebellum can “swell” depending on the size of the lesion. After the cerebellum is relaxed, the dural opening is completed in a curvilinear fashion with its base on the transverse-sigmoid junction. To improve exposure, a slit is cut toward the transverse-sigmoid junction. The dura is then tacked up anteriorly in the form of two or three triangles.

Intradurally, the junction of the tentorium and petrous bone at the superior and lateral extents of the dural opening should be visualized. The 4th through 11th cranial nerves can be identified. If necessary, the petrosal vein can be resected with impunity. Lesions of the lateral pons, lateral cerebellar hemisphere, and pontomedullary junction can be resected through this approach.

The dura is closed primarily or with a dural patch graft. Fibrin glue should be applied to the suture line to prevent leakage of CSF, and bone substitute can be used to replace the craniectomized bone. If a CSF leak develops, a lumbar drain is placed for 3 days. If the leakage has stopped at that point, the drain is removed. If the CSF leakage persists, a lumboperitoneal shunt should be considered or the patient should be returned to surgery for reclosure and rewaxing of the air cells. More recently, we have successfully applied Tisseel (Baxter Healthcare, Glendale, CA) fibrin glue to the internal auditory canal to treat CSF leaks.

Far Lateral Approach

The far lateral or transcondylar approach has several modifications and variations. It basically involves a partial condylectomy, with or without resection of the lateral mass of C1. This approach allows the surgeon to achieve an anterolateral trajectory to the brainstem, and it eliminates the need to traverse contaminated mucosal structures through the transoral or transfacial route. Thus, the risk for meningitis, CSF leakage, or both can be minimized. First described by Heros²⁶ and later modified by Spetzler and Grahm,²⁷ this approach provides access to lesions of the vertebrobasilar junction, inferolateral pons, anterolateral medulla, and upper cervical spinal cord.

Its potential disadvantages include craniocervical instability, vertebral artery injury, wound infection, meningitis, or lower cranial nerve deficits. Various positions, including the modified park bench, sitting, lateral decubitus, supine with the head rotated, and lateral or half-lateral decubitus, have been used with this approach. At our institution, we prefer a modified park bench position in which the operating table is extended 10 to 20 cm by placing a rigid plastic board under the mattress. The dependent arm is cradled in a padded sling between the table edge and the Mayfield head holder (Codman, Inc., Raynham, MA).

The clivus is brought perpendicular to the floor by performing four maneuvers on the neck: (1) flexion in the anteroposterior plane until the chin is one fingerbreadth from the clavicle, (2) rotation 45 degrees contralateral to the side of the lesion, (3) lateral flexion 30 degrees down toward the opposite shoulder (also the floor), and (4) slight distraction to increase the interval between the foramen magnum and C1 so that the surgeon can look down the axis of the brainstem and work between the horizontally oriented cranial nerves. The ipsilateral shoulder should also be retracted inferiorly with cloth tape to allow greater freedom of movement with the microscope (Fig. 394-7). SSEPs should be monitored before and after positioning to avoid inflicting a stretch injury on the brachial plexus. The patient should be taped securely to the bed to permit frequent and extreme rotation.

FIGURE 394-7 Patient position and skin incision (dashed line) for the far lateral approach. A lateral decubitus position is used with the contralateral arm cradled beyond the edge of the bed. The head is distracted rostrally and the ipsilateral shoulder is pulled caudally to increase visualization of the clivus. An inverted hockey-stick incision is used.

(Used with permission from Barrow Neurological Institute.)

Multiple skin incisions have been described, including the hockey-stick, lazy S, and simple paramedian linear incision. The traditional hockey-stick skin incision begins in the midline at C2 or C3, proceeds superiorly, and curves anteriorly and laterally to the mastoid tip. Monopolar cauterization is used to develop a plane below the superficial muscle fascia. A small muscle cuff is cut and left attached to the superior nuchal line for reapproximation of the fascia and muscles at the end of the procedure. The midline dissection proceeds down the avascular ligamentum nuchae, and the muscle flap is reflected inferolaterally with fishhooks and a Leyla bar (Aesculap, San Francisco, CA). Muscle is then stripped off the laminae of C1 and C2 with a Penfield dissector or gauze sponge. The advantage of this incision over the paramedian incision is that the dissection proceeds in a medial-to-lateral direction; hence, the vertebral artery is approached more safely.

The posterior arch of C1 is removed with a Kerrison rongeur or drill (Fig. 394-8). The lateral aspect of the ipsilateral C1 posterior arch is further removed with a rongeur to the lateral aspect of the dura. The extradural vertebral artery, which lies on top of the sulcus arteriosus, can be followed into the foramen transversarium with a Woodson or dental instrument. The venous plexus surrounding the artery can be the source of profuse bleeding, which can be controlled with Nu-Knit gauze and bipolar coagulation. The foramen transversarium can be unroofed by sliding the footplate of a Kerrison rongeur into it. Alternatively, a diamond burr can be used to skeletonize the vertebral artery. The C2 nerve root between C1 and C2 should be preserved to avoid occipital numbness. The occipitoatlantal membrane is dissected off the foramen magnum with a curved curet to prepare for the craniotomy.

FIGURE 394-8 Exposure of the lateral cerebellomedullary cistern and upper cervical cord afforded by the far lateral approach. The vertebral artery can be protected with a Penfield dissector.

(Used with permission from Barrow Neurological Institute.)

A lateral suboccipital craniotomy extending from the midline down to the foramen magnum and laterally to the retromastoid region is fashioned. If preoperative MRI shows a considerable mass effect, the craniotomy should extend across both cerebellar hemispheres so that the cerebellum has room to swell after the dura is opened. It also allows easier access to the cisterna magna. The lateral foramen magnum is drilled laterally to include the posteromedial third to half of the occipital condyle. This maneuver can be performed safely by drilling within the center of the condyle (with a diamond burr) and leaving a thin eggshell of bone that can be removed with rongeurs. The operating microscope provides superior lighting during the drilling process. The condylar emissary vein, when entered, should be expected to produce heavy venous bleeding, but it can be controlled with bone wax, cotton pads, and Nu-Knit. Visualization of this vein indicates sufficient anterior bony removal. The hypoglossal canal, which should not be seen during the exposure, is situated anterior and medial to the anterior third of the condyle.

The foramen transversarium can be unroofed with a rongeur or diamond burr, and the vertebral artery can be mobilized. The vertebral artery can be retracted medially and the lateral mass of C1 can also be drilled. The two limbs of the C-shaped dural opening are placed over the upper cervical cord and cerebellum, respectively, to increase the anterolateral exposure. The dura is tacked laterally with suture. Initially, the arachnoid is left intact to prevent extradural blood from contaminating the subarachnoid space. Once the extradural bleeding is controlled, the arachnoid is opened and tacked to the dural edge with hemoclips. The medulla, upper cervical cord, and lower cranial nerves can now be visualized.

Structures that can be accessed include the cerebellar tonsils; C1 and C2 rootlets; cerebellum; lower pons; cranial nerves IX, X, XI, and XII; the vertebral arteries; and the ipsilateral posterior inferior cerebellar artery. An extended far lateral approach, which combines a far lateral and retrosigmoid approach, provides access to the lower pons and pontomedullary junction in addition to the exposure achieved with the far lateral approach.

Supracerebellar Infratentorial Approach and Variations

In 1936, Dandy first described a supratentorial parafalcine approach to the pineal/tectal region.²⁸ He used the semisitting position and in some cases sectioned the corpus callosum, which probably damaged the deep galenic venous system and resulted in cerebral edema. He lacked the benefit of the operating microscope, MRI, frameless stereotaxy, and advanced neuroanesthetic techniques. Thus, not surprisingly, complication rates were high. As techniques evolved, it became apparent that the infratentorial route would more safely access the same area by dissecting below the galenic venous system.

This approach was initially described by Krause²⁹ and later popularized by Stein.³⁰ The supracerebellar infratentorial approach permits exposure of malformations involving the midline tectum and pineal region. Positioning is the same as for the suboccipital approach. However, the angle of the tentorium must be considered. In the ideal position the tentorium is perpendicular to the floor. This positioning can be checked preoperatively with frameless stereotactic guidance. The craniotomy should extend above and below the transverse sinus to expose the junction of the traverse sinus and torcular. This craniotomy permits greater retraction of the tentorium superiorly than possible with a pure suboccipital craniotomy. It can be performed safely by placing a single burr hole lateral to the superior sagittal sinus with a pneumatic drill, footplate, and drill bit. Before the sinus is crossed, the surgeon should reverse the footplate and irrigate through the craniotomy line to confirm that the plate is extradural.

The dura is opened in an inverted V shape with the base on the edge of the transverse sinuses (Fig. 394-9). Bridging veins from the superior aspect of the cerebellum that drain into the transverse sinus are coagulated and divided to permit downward retraction of the cerebellum. If these veins are not coagulated early in the procedure, severe venous bleeding can occur during retraction. They must be coagulated as close to the surface of the cerebellum as possible to leave a pedicle on the surface of the tentorium. If avulsed at the interface of the tentorium, coagulation is not only futile but may enlarge the hole in the sinus and cause disastrous bleeding. If this occurs, a piece of Nu-Knit larger than the defect should be patched over the hole. A cotton pad is then placed over the hemostatic agent and can be held in place with a retractor during the remainder of the procedure. At the end of the procedure, the retractor is removed and the Nu-Knit will adhere to the venous opening.

FIGURE 394-9 The craniotomy (black dashed line) for the supracerebellar infratentorial approach extends across all three sinuses so that when the dura is opened (green dashed line), the tentorium can be retracted superiorly without obstructing bone over the torcular.

(Used with permission from Barrow Neurological Institute.)

During the approach to the deep galenic system, the arachnoid is often thick and should be divided over the veins as close to the cerebellum as possible. Care must be taken to identify the vein of Galen, internal cerebral veins, basal veins of Rosenthal, occipital veins, pineal veins, and precentral cerebellar vein. If possible, the latter should be preserved; if not, it can usually be coagulated and divided with impunity. Further exposure can be achieved by incising the tentorium 1 cm lateral to the straight sinus (see the supracerebellar transtentorial approach below). The upper vermis can also be resected to explore the lower posterior midbrain and anterior medullary velum.

The CM is dissected with the usual set of instruments. Long bipolar coagulation devices and microdissectors, including down-biting, up-biting, and straight microdissectors, should be used. Long suction instruments, ultrasonic aspirators, pituitary rongeurs, and tumor forceps are also useful. The resection bed should be inspected for residual lesion and venous anomalies. The dura should be closed in watertight fashion with a dural patch and fibrin glue, if necessary.

This approach restricts maneuvers to below the deep galenic system and avoids traversing brain tissue. No normal tissue is violated if the lesion is ventral to the velum interpositum and deep venous system. The limitations of the approach are defined by the superolateral extension of lesions above the tentorium, which can be difficult to reach from an infratentorial exposure.

Paramedian and lateral (retrosigmoid) supracerebellar infratentorial approaches have also been described. The former is performed through a paramedian craniotomy 2 to 3 cm off midline and provides direct access to the cerebellomesencephalic fissure, inferior colliculus, fourth cranial nerve, and superior cerebellar peduncle. The latter approach is through a retrosigmoid infratentorial craniotomy and provides access to lesions of the lateral quadrigeminal plate, superior cerebellar peduncles, trigeminal nerves, posterolateral mesencephalon, and ambient cistern. In addition to providing a more direct surgical corridor to these locations, the paramedian and lateral variations of the supracerebellar infratentorial approach also avoid the majority of tentorial bridging veins, which are clustered more in the midline.

In the supracerebellar transtentorial variation of this approach, the tentorium is incised to expose the mesial and inferior cortical surfaces of the temporal lobe and posterolateral rostral mesencephalon.

Interhemispheric Transcallosal Approach

Deep-seated supratentorial lesions, such as CMs of the thalamus, lateral ventricle, third ventricle, and corpus callosum, can be approached via either an anterior or posterior interhemispheric transcallosal approach. For periventricular lesions, the surgical corridor can be determined by using a variation of the two-point method in which an ependymal surface is substituted for a pial surface. An improved trajectory to the lesion is often obtained by approaching the lesion from the contralateral side (i.e., contralateral interhemispheric transcallosal approach). The benefit from approaching contralaterally is increased the more lateral the lesion is situated within the ventricular system.

Patients are positioned supine, and a rolled towel or other support is placed under their contralateral shoulder so that the head is turned 90 degrees ipsilaterally (the long axis of the sagittal sinus thus being brought parallel to the floor). The patient’s head is flexed laterally with the vertex up 35 to 45 degrees. For the anterior interhemispheric approach, it is helpful to extend the patient’s neck so that the angle of attack can be as anterior as possible. Frameless stereotactic guidance is essential. In general, the further anterior the trajectory, the less chance of encountering bridging veins. Similarly, for posterior interhemispheric approaches, the head should be flexed and the craniotomy centered over the posterior third of the sagittal sinus.

A linear incision oriented in the coronal plane is centered on the planned craniotomy location. Classically, for anterior interhemispheric approaches, the craniotomy is placed so that two thirds of the rectangular bone flap is situated anterior to the coronal suture and one third is posterior. The bone flap is extended approximately 1 cm across the sagittal sinus on the side contralateral to dissection of the interhemispheric fissure. The sagittal sinus can then be gently retracted from the field of view. Usually, 3 cm of bone exposure on the ipsilateral side is more than sufficient. In practice, the exact position of the bone flap is best determined in conjunction with image guidance because prominent bridging veins can easily be seen and the bony opening adjusted accordingly. This is equally true for the posterior and anterior interhemispheric approaches.

A burr hole can be placed either next to the sagittal sinus or directly on the sinus by using a pneumatic drill with a fluted cutting burr at both the anterior and posterior margins of the craniotomy. A side-cutting drill bit with a footplate is then used to elevate the flap. Nu-Knit may be placed on top of the exposed sinus to control epidural oozing. Once the dura is exposed, circumferential epidural tack-up sutures are placed and the dura is incised with a No. 15 blade. The dura is then reflected carefully over the sagittal sinus and tacked with 4-0 Nurolon suture. Care must be taken because some bridging veins may enter the dura or venous lakes in the dura over the cerebral convexity. These veins must not be sacrificed. If a vein cannot be dissected free, it can usually be isolated as it courses in the dura.

Next, dissection of the interhemispheric fissure is performed. The dependent cerebral hemisphere requires no retraction because the force of gravity itself is sufficient. This dissection is usually straightforward. The surgeon, however, must be aware of dehiscent falx cerebri and avoid vascular injury to the anterior cerebral arteries or their branches. The corpus callosum is readily identified by its pallor and is entered through a minimal opening, which is then gently stretched parallel to its long axis.

Once the ventricle is entered, further steps depend on the location of the lesion. CMs located within the ventricle per se can be removed safely, usually en bloc. These lesions may have significant venous components that may be closely associated with choroidal veins. Utmost care in preserving these veins and establishing hemostasis is critical. A ventricular drain may be placed as a precaution if the surgeon is concerned about the possibility of postoperative hemorrhage.

The choroidal fissure must be opened to expose lesions in the third ventricle. For thalamic or other basal ganglia lesions extending into the lateral ventricle, the same surgical considerations extended to brainstem CMs apply (i.e., reliance on image guidance to plan the minimally traumatic trajectory to the lesion, a minimal opening, and careful piecemeal microdissection). In our experience, the pulvinar is the most forgiving entry zone for intrinsic thalamic lesions. It can even be used to approach intrinsic lesions of the basal ganglia that cannot be accessed from an anterior medial approach because of the risk to the internal capsule.