Anatomical Hemispherectomy

Gregory G. Heuer and Phillip B. Storm

The general principle of surgery for intractable epilepsy is the removal or isolation of the region of brain that is generating the seizures from the normal brain. Anatomical hemispherectomy is epilepsy surgery in its most radical form, namely the surgical removal of the entire hemisphere.

Background

Walter Dandy is credited with performing the first hemispherectomy.1 In 1929, Dandy performed a resection of the entire hemisphere in a patient with a diffuse right-sided glioma. The patient survived the procedure, and Dandy felt the patient did have some clinical benefit. Despite the early clinical benefit, the patient eventually died as the glioma recurred. The first hemispherectomy for epilepsy was performed by McKenzie in 1938.² The patient was cured of the epilepsy after the procedure. The first clinical series of the use of hemispherectomy to treat epilepsy was published by Krynauw in 1950.³ He described a series of 12 patients, mostly with infantile hemiplegia syndrome, and found most of the patients, 10 of 12, had a significant reduction in seizures after surgery.

Anatomical hemispherectomy is most often performed in the young pediatric population. Indications for hemispherectomy include conditions in which the pathological process wholly or in large part affects one entire cerebral hemisphere. These conditions include Sturge-Weber syndrome, Rusmussen’s encephalitis, porencephaly syndromes, schizencephaly, trauma, hemimegalencephaly, and cortical dysplasia involving a significant area of a single cerebral hemisphere.

The seizure control obtained in these early studies was encouraging, and the technique was widely employed in the 1960s. After a large number of the surgeries were performed, some common complications were found to be associated with the procedure. These complications included late postoperative hydrocephalus and superficial cerebral hemosiderosis.⁴ These complications have been described to occur in up to a third of patients in the early studies.^5–7 Some technical modifications of the procedure have led to decreased complications and maintained the seizure reduction seen in the early studies.^8–12 More recent series have demonstrated seizure control in 60 to 94% of patients after hemispherectomy.^12,13 The seizure control rates are partial dependant on the age of the patient at the time of surgery, the specific disease pathology, and the length of time that seizures have been occurring.

Operative Technique

The availability of a specialized pediatric anesthesiologist familiar with the hemispherectomy procedure and the unique associated risks is imperative. A major risk of an anatomical hemispherectomy is blood loss. Prior to beginning the procedure it is imperative to have excellent intravenous (IV) access and blood products available. A pulmonary artery catheter in not routinely used, but central lines and arterial lines are placed in all patients. Frequent dialogue between the neurosurgeon and the anesthesiologist is imperative to accurately record and follow blood loss. When blood loss reaches one blood volume, we seriously consider aborting the procedure, especially if there is a substantial amount of brain that needs to be resected along the sagittal or transverse sinus where there is potential for brisk bleeding.

After the patient is anesthetized and adequate monitoring and venous access is obtained, the patient is positioned in the supine position with a gel roll under the ipsilateral shoulder and hip. The head is supported in a Mayfield horseshoe headholder (Integra Lifesciences Corp., Plainsboro, NJ) and rotated so that the nose is parallel with the floor. If the neck is not supple or there is a concern of decreased venous drainage, the bed is rotated until the head position is ideal. The head of the bed is raised to ~30 degrees. The hair is minimally shaved with electric clippers. Prior to making the incision, the patient is given antibiotics and dexamethasone. However, unlike tumor resections or other epilepsy surgeries, mannitol is not given.

A T-shaped incision is used. The first incision is made in the sagittal plane, starting in the midline behind the hairline and extending posteriorly until the inion. A second incision is made in the coronal plane, starting at the level of the zygoma in front of the ear and extending so that it intersects the initial incision in the midline at a right angle (Fig. 14.1A). The scalp is then reflected to give excellent exposure of the skull (Fig. 14.1B). During the exposure, steps are taken to reduce blood loss including infiltrating the wound with a solution of bupivicaine and epinephrine and the placement of Raney clips on the wound edges.

A series of bur holes are placed in the midline directly over the sagittal sinus. Multiple bur holes are placed to decrease the risk of injuring the transverse and sagittal sinuses. In children <18 months of age, the dura is often difficult to free from the overlying bone. Therefore, bur holes are placed within 2.5 cm of each other along the sagittal suture. Additional bur holes are placed in the keyhole and above the zygoma in the squamous temporal bone. A craniotome is used between the bur holes. If necessary, the bone flap is extended with a Leksell rongeur and Kerirson punches so that the sagittal sinus and the torcular can be adequately visualized. It should be noted that occasionally the craniotomy can result in copious blood loss to a degree that necessitates aborting the procedure. This is more common in very young patients. In those instances in which a significant blood loss occurred with the craniotomy and the procedure had to be aborted, the patient returns to the operating room in 1 to 2 weeks and the hemispherectomy is performed.

Fig. 14.1 Serial surgical photographs from a representative left hemispherectomy. (Note: in all figures medial is at the bottom and anterior is at the left.) (A) Representative incision. (B) Surgical field after the skin incision. (C) Radial cuts made toward the sagittal sinus. (D) Postresection surgical field.

Once the bone is removed, the dura is opened in a C-shaped fashion based on the sphenoid bone. The opening should be at least 4 cm lateral of the sagittal and transverse sinuses to avoid bridging veins. Approximately four to eight radial cuts are made toward the sagittal and transverse sinuses (Fig. 14.1C). The cuts are easily tailored anteriorly or posteriorly to avoid bridging veins. If the bridging veins cannot be spared, the portion of dura attached to the vein is left intact and the vein is coagulated after arterial feeders are removed along with adjacent brain.

Prior to beginning the resection, some specialized devices and materials are added to the surgical field. The simultaneous use of two bipolar setups enables two surgeons to work independently and can help to reduce the total blood loss. The use of Isocool Bipolar Forceps (Codman, Raynham, MA), designed to reduce tissue sticking to the bipolar tips, may be helpful in large resections. Other adjuncts, such as thrombin-soaked cotton balls or Floseal (Baxter, Deerfield, IL) can minimize blood loss.

The resection is performed in a step-wise fashion based on anatomical regions. First the middle temporal gyrus is entered and the dissection is carried down to the temporal horn of the ventricle. Next, the dissection is continued in the middle temporal gyrus approximately 2 cm deep to the gray-white junction, and continued posteriorly in a C-shaped fashion around the sylvian fissure. This dissection plane is then carried anteriorly along the inferior frontal gyrus. Duringthis part of the dissection, the branches of the middle cerebral artery (MCA) are coagulated. The result of this dissection is an extended C-shaped trough from the temporal lobe, behind the sylvian fissure, and along the inferior frontal lobe. The trough is packed with thrombin-soaked cotton balls. After meticulous hemostasis is achieved, the temporal lobe and the frontal lobes are removed simultaneously.

The temporal lobectomy is performed in the standard fashion with the following exceptions. After the standard anterior lobectomy, the entire superior gyrus is removed starting at the most proximal portion of the sylvian fissure and moving posteriorly until the initial trough is encountered. The vein of Labbé is spared at this point unless it is very small or is under tension and therefore at risk for inadvertent rupture. The distal MCA branches are coagulated and cut as they are encountered. Weck clips and aneurysm clips are available but are normally unnecessary if the arteries are adequately coagulated. An amygdalohippocampectomy is then performed.

After the temporal lobe is resected, the frontal lobe dissection is performed. If two surgeons are available, it is possible to perform the dissection of the frontal lobe simultaneously with the removal of the temporal lobe. The dissection is begun in the motor strip at the level of the initial trough, and it is carried around the sylvian fissure and then medially toward the sagittal sinus. The dissection is continued anteriorly in the superior frontal gyrus until the orbital ridge is encountered. Bridging veins from the superior frontal gyrus are left intact. As before, the trough is packed with thrombin-soaked cotton balls. After meticulous hemostasis is achieved, the superior, middle, and inferior frontal gyri are undermined approximately 1 cm deep to the gray-white junction and removed in one large piece.

Next, the bridging veins are coagulated that arise from the remaining superior frontal gyrus, moving in an anterior to posterior direction. Because it is difficult to control bleeding adjacent to the sagittal sinus, these veins should be cut close to the cortex and not close to the sinus. After coagulating and cutting the bridging veins the remaining portion of the superior frontal gyrus is retracted laterally off of the falx. The cingulate gyrus, anterior cerebral arteries (ACA), and corpus callosum are identified. It is very important to remember that the branches of the ACA are often stacked in a superior inferior orientation and not right and left. Great care must be taken not to incorrectly identify these arteries and accidentally sacrifice the contralateral artery. Preoperative imaging, such as an angiogram or magnetic resonance imaging (MRI), may be useful in defining the relationship of the vessels. Using a subpial dissection technique to remove the cingulate gyrus, using gentle suction, and refraining from the use of the bipolar cautery can reduced the risk of damaging the ACA branches. Staying lateral to the pial layer, the corpus callosum is entered and split with the bipolar on a setting of 20 J and with gentle suction. This dissection is then carried posteriorly to complete the callostomy. During the dissection, once the ipsilateral ACA is correctly identified, it may be clipped proximal to the origin of the callosomarginal.

The parietal lobe is then removed similarly to the frontal lobe, by starting laterally at the level of the initial trough around the sylvian fissure and working posteromedially in the parietal gyrus adjacent to the parietooccipital sulcus. The dissection is then continued anteriorly, parallel to the sagittal sinus, again avoiding bridging veins. The parietal lobe is then undermined apporoximately 1 cm below the gray-white junction. As before, the bridging veins going into the sagittal sinus are coagulated. The remaining portion of the parietal lobe is gently lifted off of the falx, coagulating any bridging veins as it is lifted. The splenium of the corpus callosum is identified along with the pericallosal artery. The remaining portion of the parietal lobe is removed above the cingulated gyrus. We again perform a subpial removal of the cingulate gyrus with suction and avoid the use of the bipolar.

The last lobe removed is the occipital lobe. Before removing the occipital lobe, examine the bridging veins going into the transverse sinus with particular attention to the location of the vein of Labbé. The occipital lobe at this point is quite mobile, and the surgeon must be careful not to avulse a bridging vein out of the sagittal, transverse, or sigmoid sinus. We typically coagulate these veins prior to removing the occipital lobe. Once the veins are cut, the occipital lobe can easily be manipulated. However, the bridging veins along the tentorium must be coagulated prior to lifting up the occipital lobe. This is started at the most posterior location of the temporal lobectomy in the lateral ventricle. We then bipolar and aspirate inferiorly to the tentorium. The brain-stem is located medially, traversing the incisura. Care should be taken to stay in the subpial plane to avoid coagulating perforators in the ambien cistern. The posterior cerebral artery (PCA) can be safely coagulated after it exits the ambien cistern. Next, the dissection is continued posterior-medially, lifting the occipital lobe off the tentorium and coagulating bridging veins. A useful anatomical reference is the occipital horn of the ventricle, as it serves as a marker for the appropriate depth of resection. The dissection is continued just superficial to the ependyma as the falx is approached. Prior to the final disconnection, the area along the falx is examined by retracting the occipital lobe laterally, to assure that there are no veins or arteries that have not been coagulated. The occipital pole usually lifts out very easily. However, during the process of removing the occipital lobe, one should be aware of any veins that are tethering this tissue to a venous sinus or the tentorium.

If adequate hemostasis is achieved when removing the individual lobes, the final inspection of the surgical cavity normally demonstrates very little residual bleeding (Fig. 14.1D). The exception is in cases were the total blood loss was excessive or where the surgery took an excessive amount of time. In these instances, peri- and postoperative coagulopathiescan occur, resulting in life-threatening bleeding.14^,¹⁵ Prior to closing, the MCA, PCA, and ACA should be inspected to assure that these vessels are adequately coagulated.

After meticulous hemostasis is obtained, the surgical cavity is copiously irrigated to remove debris. The removal of this debris possibly reduces the development of the postoperative fever that has been associated with this material.¹⁶ The dura is closed in a water-tight fashion and the closure is augmented with an onlay dural substitute such as DuraGen (Integra Lifesciences Corp., Plainsboro NJ). The placement of dural tacking sutures may be useful in preventing the development of epidural collections. The bone is attached with Synthes absorbable plates and screws (Synthes, West Chester, PA), and the incision is closed in the standard layered fashion. A comparison of preoperative (Fig. 14.2A) and postoperative (Fig. 14.2B) imaging can confirm a proper resection.

Postoperatively, patients are observed in the intensive care unit. For the first 48 hours, we keep the patient positioned with the craniotomy side up to give the newly created cavity time to fill up with CSF. This allows the pressure to equalize in the two hemispheric compartments and may reduce the risk of life-threatening brain shift.¹⁷ After the 48 hours, patients begin rigorous physical therapy and occupational therapy. The patients are watched carefully for signs and symptoms of hydrocephalus. Within the first 2 to 3 weeks after surgery the diagnosis can easily be confirmed by the prescence of a large, firm pseudomeningocele. After a few weeks, the diagnosis may be more subtle, therefore, we obtain a baseline computed tomography (CT) prior to discharge. An important technical point is to place the proximal catheter in the contralateral ventricle, not the resection cavity. This substantially decreases the formation of a pseudomenigocele and a cerebrospinal (CSF) leak because the brain on the contralateral side will seal the tract around the catheter.

Complications

The postoperative complications that occur with any neurosurgical procedure, such as wound infections and strokes, can occur after hemipherectomies. However, there are some postoperative complications that are unique to this procedure. Although extremely rare, significant brain shift can occur and result in sudden death after a hemispherectomy.¹⁷ Aspetic meningitis can occur in the postoperative period and is often effectively treated with a short course of steroids.¹⁸^,¹⁹ As stated earlier, irrigation of the surgical cavity and removal of any debris may reduce the incidence of this complication. Hydrocephalus can develop in patients at two time periods: in the immediate postoperative period and in a delayed fashion in some patients.^19,20–25 This must be recognized and treated with standard shunt placement.

Conclusion

Anatomical hemispherectomy is an important and useful surgical technique in selected patients with intractable epilepsy. A complete understanding of the surgical anatomy is needed to safely perform this procedure and care must be taken to monitor and minimize blood loss. When performed carefully in the correct patients at an institution capable of handling the unique surgical and postoperative needs of these patients, hemispherectomy can result in significant control of a patient’s seizures.