Historical Background

The standard temporal lobectomy developed concurrently with the identification of the temporal lobe epilepsy syndrome and the emergence of electroencephalography (EEG) in the 1940s and 1950s. Much of the credit belongs to Penfield and Jasper at the Montreal Neurological Institute (MNI).¹ When Herbert Jasper joined the MNI in 1937, he brought the technique of EEG with him. By this time, Penfield and his partner William Cone had been performing surgery for epilepsy since Penfield arrived in 1928. As EEG became more established in the evaluation and diagnosis of focal epilepsy through the work of Jasper and Gibbs, the classification and study of temporal lobe epilepsies was under way. Through the close friendship of Jasper and Penfield, the MNI became a leader in the late 1930s in the surgical treatment of epilepsy. In 1941, Jasper and Kershman proposed a classification of epilepsy based on EEG waves.² In this report, they described the localization of psychomotor phenomena from within the deep regions of the temporal lobe. Despite increasing awareness of the role of the mesial temporal lobe in these seizures, the lack of understanding about the function of this tissue and the inability of the surgeon to “see” the lesion at the time of resection led to reluctance on Penfield’s part to remove these structures.¹ In 1950, Penfield reported his success in anterolateral temporal resections.³ In this series of 68 patients, 10 had partial removal of the uncus, and only 2 had hippocampal resections. EEG abnormalities were recorded from the temporal lobe in this group of patients, but an underlying substrate had not been identified. At the same time, Percival Bailey and Ernest Gibbs reported a series of 25 patients from the University of Illinois program who underwent temporal resection guided by EEG.⁴ Similar to Penfield’s series, these patients did not have hippocampal resections.

The second phase in the development of the modern surgical strategy to treat temporal lobe seizures took place in the 1950s as the role of the mesial temporal lobe structures in the pathogenesis of the epilepsy became better understood. This occurred as a number of scientists began to study the connections of the mesial temporal lobe to the rest of the brain and through the use of stimulation studies to reproduce seizure semiology in animals and humans.^5–8 In 1952, Penfield and Baldwin published a classic monograph describing their technique for anterolateral temporal lobectomy including the hippocampus and amygdala.⁹ They reported that the most frequent pathologic abnormality in two of three of their cases was an atrophic lesion termed incisural sclerosis. Falconer in a report in 1953 recognized a connection with febrile seizures and introduced a modification allowing for en bloc resection of the hippocampus, which allowed the pathologists to study the tissue.¹⁰ With the advent of neuroimaging, modifications to the technique reported by Penfield have been made to address specific pathology seen preoperatively. Despite these modifications, the operation developed and introduced by innovative neurosurgeons in the early 20th century remains one of the most successful operations for the treatment of epilepsy today.

Identification of Surgical Candidates: The Concept of Pharmacoresistance and Medical Intractability

The indications for epilepsy surgery generally include the presence of focal epilepsy resistant to treatment with an adequate trial of anticonvulsant therapy. The precise definition of an adequate anticonvulsant trial is open to interpretation, but a study by Kwan and Brodie produced useful information regarding the efficacy of anticonvulsant therapy in newly diagnosed epilepsy.¹¹ This study suggests that after three medications fail to control seizures, further success is unlikely, and other options should be considered. Consider also the paper by Wiebe and associates, comparing temporal lobectomy to optimal medical therapy in a group of patients with temporal lobe epilepsy.¹² In this prospective randomized trial, surgical therapy in combination with medical therapy was far superior to ongoing medical therapy alone. These papers lend credence to the idea that patients with ongoing epilepsy despite a trial with a few anticonvulsants should be expeditiously evaluated for possible epilepsy surgery.

Preoperative Evaluation

Surgical Decision Making

For the TLE patient, decisions revolve around localizing the epilepsy to the lateral or mesial temporal lobe, the presence and nature of the epileptic lesion, and the presence of language or important short-term memory deficits. For the patient undergoing a standard temporal lobectomy, the epilepsy should be localized to the anteromesial temporal lobe, and ideally a well-defined lesion should be present (mesial temporal sclerosis, malformation of cortical development, neoplasm, cavernoma). The planned posterior extent of resection should not encroach on possible neocortical temporal lobe language areas. If the dominant temporal lobe is involved, a baseline memory or naming deficit in the presence of mesial temporal sclerosis would support the conclusion that the correct brain site was targeted, and the risk for causing further neurological deficits would be acceptably low.^27,28

The anatomy of the temporal lobe deserves a brief discussion at this point. The temporal lobe has well-defined anterior, lateral, basal, and mesial surfaces. The posterior boundary is arbitrary, having no obvious anatomic demarcation separating it from the parietal area. The temporal lobe is made up of five gyri and their corresponding sulci. The lateral surface lies below the sylvian fissure and extends to the floor of the middle cranial fossa. The gyri from top to bottom include the superior temporal gyrus (T1), the middle temporal gyrus (T2), and the inferior temporal gyrus (T3), which often extends onto the basal surface. The basal surface includes the inferior temporal gyrus (T3), the fusiform gyrus, and the parahippocampal gyrus. The mesial surface includes the amygdala and the parahippocampal gyrus, including the uncus. The collateral sulcus separates the fusiform and parahippocampal gyri and serves as an important reference to locate the temporal horn of the ventricle. Within the temporal horn, important anatomic structures include the inferior choroidal point (anterior choroidal artery enters the choroid plexus here), the hippocampus occupying the mesiobasal portion of the ventricle, the fornix, the choroid plexus, the choroidal fissure, and the amygdala in the anterior-superior-medial portion of the ventricle. The reader is referred to an excellent series of articles describing in detail the temporal lobe anatomy.²⁹

Functional anatomy in the temporal lobe includes comprehensive language cortex in the dominant temporal lobe, visual field fibers (Meyer’s loop) subserving the contralateral upper quadrantic visual field information, and potential important memory and naming centers. The anatomy of language cortex in the temporal lobe can be quite variable.^30,31 Cortical stimulation testing can be performed to further identify and protect lateral language cortex when posterior temporal lobe resections are anticipated on the dominant side. This can be done either intraoperatively with the patient awake or extraoperatively with implanted electrodes. After the language cortex has been identified, the temporal resection can be tailored to the patient’s individual anatomy. The standard temporal lobectomy is designed to avoid temporal lobe cortical language sites by limiting the resection of the superior temporal gyrus to 3 to 4.5 cm from the anterior temporal pole. There is some controversy as to whether even this practice is safe, and some centers advocate leaving the entire superior temporal gyrus in place. This is based on language-stimulation data suggesting the presence of language sites in the anterior 3 cm of the superior temporal gyrus in a small percentage of patients.³⁰ Of course, these sites were not resected, and thus it is difficult to know whether they were essential language sites. Other “nonessential” language sites in the temporal lobe have been demonstrated through cortical stimulation followed by resection.³² These sites were located in the basal temporal lobe by direct cortical stimulation but left no permanent language deficit after they were resected. Insufficient data, however, are available from the limited numbers of patients reported to conclude with certainty that no patient will develop a permanent language deficit after resection of a basal temporal language site.

Visual field fibers are also located in the temporal lobe as they extend forward from the lateral geniculate body before turning posterior on their way to calcarine cortex. These fibers are located unpredictably in the roof of the temporal horn, and standard temporal lobe resections cause injury to this fiber tract in as many as 50% of cases.³³ This leads to the “pie in the sky” visual field deficit with loss of peripheral vision in the opposite upper quadrant. In most cases, the visual field defect noted with careful perimetry testing is not clinically significant.

Finally, other important anatomic structures the surgeon should be familiar with include the sylvian fissure and associated structures (sylvian vein, middle cerebral arterial cascade, and underlying insula), the vein of Labbé, and the region of the tentorial incisura, including the brainstem, posterior cerebral artery, basal vein of Rosenthal, and third and fourth cranial nerves. Familiarity with these anatomic structures is critical to avoid a potentially devastating injury during resection of the mesial temporal lobe structures.

Surgical Technique

When the decision has been made to perform a standard temporal lobectomy, the technique varies little from patient to patient. The word standard implies a reproducible operation from patient to patient and even surgeon to surgeon. It is best used to treat the “standard” syndrome of mesial TLE consisting of reproducible EEG, MRI, and seizure semiology elements. The targets of this operation are the mesial temporal lobe structures that are “sclerotic”: the parahippocampus, hippocampus, and amygdala. The exact posterior extent of resection of the hippocampus to improve outcomes is unknown, but at least one prospective trial correlated improved outcome with more aggressive resection of the hippocampus.³⁴

The procedure does vary slightly for the language-nondominant and language-dominant sides. On the dominant side, the surgeon limits resection of the superior temporal gyrus to avoid a possible postoperative language deficit. This usually translates into a resection of 3 to 4.5 cm measured along the superior temporal gyrus from the anterior aspect of the middle fossa. On the nondominant side, the posterior extent of resection can be farther from the temporal pole, although it is usually limited by the vein of Labbé about 4.5 to 6 cm posterior to the temporal pole. One must be careful with resections extending posteriorly beyond these measurements because there is increased risk for injuring the geniculocalcarine tract with resultant homonymous hemianopsia.

Preoperatively, the patient is given antibiotics within 1 hour before incision. A general anesthetic is administered, and bladder and arterial cannulations are performed. Intravenous access is maintained throughout the case with peripheral catheters. Before incision, hyperventilation therapy and intravenous mannitol can be used to relax the brain and minimize retraction during surgery. The patient is positioned supine on the operating table with the head rigidly fixated by a head clamp attached directly to the operating table. Stereotactic navigation is optional and is not routinely used at our institution. The position of the head is important because optimal positioning allows the surgeon to access the mesial structures with less retraction on the temporal lobe. Optimal positioning includes placing an ipsilateral shoulder roll to minimize torsion on the neck and then turning the head 30 degrees from the midline so that the operative side is up. The head is slightly extended to bring the sylvian fissure to a perpendicular plane to the operating approach. Finally, dropping the vertex down toward the floor allows the surgeon easier access to the mesial structures and allows less retraction on the temporal lobe.

Once positioned, the hair in the frontotemporal region is clipped, and a “reverse question mark” incision is made from just above the zygoma, extending back in the temporal region to the posterior part of the pinna and then curving anteriorly just above the insertion of the temporalis muscle. A larger skin flap is not necessary and may lead to increased risk for cosmetic deformity. The skin incision is carried out staying in the plane above the temporalis fascia. I (W.B.) prefer to split the fascia of the temporalis muscle and then elevate the muscle using a T incision based on the zygoma, with most of the muscle bulk reflected anteriorly.³⁵ This approach is done with a scalpel to minimize shrinking of the muscle fibers and facilitates an easier reattachment of the muscle at the end of the procedure. One should also be careful to leave enough muscle cuff attached to the temporal bone to allow secure suturing of the muscle at closure. Despite all these efforts, a significant cosmetic deformity may occur from wasting of the temporalis muscle, and this should be discussed with the patient before surgery. At this stage, exposure of the temporal bone from the root of the zygoma to the anatomic “keyhole” should be visualized. The anterior aspect of the temporalis muscle is undermined with electrocautery in case the bone in the region of the sphenoid wing needs to be rongeured away to allow additional exposure of the temporal pole. In most cases, this additional removal of bone is unnecessary and further increases the chances of a cosmetic deformity after surgery.

The craniotomy should facilitate exposure of the lateral aspect of the temporal lobe from the base of the middle fossa to the sylvian fissure. In our practice, the frontal lobe is not exposed during this procedure, but other epilepsy surgeons do routinely expose the posterior-inferior aspect of the frontal lobe during this operation. The anterior aspect of the bone removal should extend to the sphenoid wing, and the spine of the sphenoid bone is removed with a fine rongeur. The craniotomy can be performed with two bur holes, respectively located at the base of the zygoma and the keyhole. The bone is then removed with high-speed drilling, and the final break across the sphenoid is performed after removing the outer cortex with the drill or fine rongeur. All bone edges should be waxed as necessary to stem bleeding, and any exposed air cells along the temporal base are sealed. Restricting the craniotomy to the bone below the temporalis muscle cuff allows placement of the titanium fixation plates entirely below the muscle, which prevents the patient from feeling them through the scalp after surgery.

At this point, the dura is inspected, and bleeding is controlled. The dural opening should be created to maintain some blood flow into the dural flap. This is best accomplished by reflecting the flap anteriorly and inferiorly so that the middle meningeal branches are maintained. The dural flap is reflected, and the temporal lobe is visualized. The sylvian fissure is recognizable along the superior limit of the temporal lobe, and the floor of the middle fossa should be visualized with minimal retraction of the inferior temporal gyrus. Additionally, as the surgeon looks anteriorly along the sylvian fissure, the anterior extent of the temporal pole should be visualized within 1 to 2 cm of the anterior bony edge of the craniotomy. If not, additional bone is removed as discussed earlier. Also at this stage, brain swelling should be assessed and changes in the anesthetic technique made if necessary. Head elevation can be a useful maneuver to reduce intracranial swelling.

The posterior limit of resection along the superior temporal gyrus is now measured with a Penfield dissector placed so that the curve of the instrument follows the curve of the temporal pole and the tip of the instrument contacts the dura of the anterior middle cranial fossa. The appropriate distance is chosen based on the side of surgery and the location of important draining veins and arterial branches supplying the posterior temporal lobe. The lateral cortical resection is designed to allow access to the deeper mesial structures; therefore, preservation of the veins and arteries supplying the posterior temporal cortex is extremely important. It is possible to achieve the goals of surgery with a small lateral resection but is difficult to reverse neurological deficits associated with a larger resection that damages the blood supply to brain tissue that is not resected. It is wise to attempt to preserve all draining veins that connect to the sylvian venous system or to the vein of Labbé. Smaller veins draining to the anterobasal dura of the middle fossa (temporal tip veins) can be ligated. Assessment of the venous drainage pattern of the frontotemporal region should be performed before the start of the lateral resection because the sylvian venous system is often variable. Care should be exercised as one dissects along the anterior-superior temporal gyrus and temporal pole to avoid injury to the outflow of the sylvian vein where it enters the sphenoparietal sinus. If significant variations in the venous pattern exist, modifications of the lateral resection should be attempted to minimize disruption to these veins. In the worst-case scenario, the operation can be converted to a selective transcortical or transsulcal amygdalohippocampectomy (when removal of the mesial temporal structures is the goal of the operation).

After the posterior line of resection is marked, dissection begins along the superior temporal gyrus a few millimeters inferior to the sylvian fissure. This is done with bipolar coagulation and sharp dissection of the pia mater followed by subpial aspiration of the cortical tissue. This allows exposure of the temporal pia of the sylvian fissure and the underlying insula and middle cerebral artery. The cortex should be aspirated down to the level of the inferior circular sulcus of the insula, where the pia ends and the temporal white matter begins. This marks the depth of the initial lateral resection to avoid injury to deeper structures. This dissection should extend anteriorly along the pia of the superior temporal pole until the dura of the anterior aspect of the middle fossa is reached. The posterior extent continues to the premeasured point determined by the side of surgery (3 to 4.5 cm left, 5.5 to 6 cm right). The posterior line of resection extends from just below the sylvian fissure at the superior temporal gyrus angling posteriorly along the middle and inferior temporal gyrus so that slightly more inferior temporal gyrus is removed than superior temporal gyrus. The pia along the superior and inferior temporal sulci is coagulated and divided during this phase of the procedure. The cortical tissue is aspirated down to the depth determined by exposure of the inferior circular sulcus. The basal temporal lobe is divided along the line of the posterior cut as the fusiform gyrus is aspirated to expose the collateral sulcus. This can sometimes be confusing, and it is important to retain orientation during this stage so as not to injure the deeper midline structures. This can be avoided by ensuring that the dura of the middle cranial fossa is still beneath the pia being dissected and also by gently retracting the basal temporal tissue and looking for the edge of the tentorium. If the edge of the tentorium is encountered, it is likely that the collateral sulcus has already been divided. The collateral sulcus is an important landmark because it facilitates controlled entry into the inferior horn of the lateral ventricle. This entry point is designed to be on the inferior-lateral aspect of the temporal horn to avoid potential injury to the optic radiations. The ventricle is identified by gently aspirating the white matter just distal to the end of the collateral sulcus, and cottonoid patties can be placed to keep the ventricle open and protect the choroid plexus. The lateral aspect of the ventricle can then be opened and the white matter of the anterior temporal lobe divided in a line connecting the lateral opening of the ventricle and the inferior circular sulcus of the insula. As the dissection progresses, the intraventricular surfaces of the amygdala and hippocampus become apparent. The final disconnection of the lateral temporal lobe occurs as a cut through the lateral ventricular sulcus (collateral eminence) is made and the basal temporal pia divided just deep to the collateral sulcus. This pia is then divided anteriorly to the prior cut made from above and ending at the temporal pole. With this, the lateral temporal lobe is free and can be removed after inspecting to ensure all draining veins to the dura have been coagulated and divided.

The next stage of the operation can be performed with loupe magnification or the operating microscope. The operating microscope has the advantage of magnification and illumination and is recommended. The anatomy in this region is complex, and careful removal of the parahippocampus, hippocampus, and amygdala requires a thorough understanding of the relationships existing among the structures in the perimesencephalic cistern, the hippocampal sulcus, and the choroidal fissure and point. These relationships are better visualized with the operating microscope. At this point, the mesial resection can be thought of in two stages, with either stage proceeding first. These consist of the amygdalar-uncal removal and the hippocampal-parahippocampal removal. These stages are performed using the subpial aspiration technique, the one exception being division of the superior aspect of the amygdala in a line connecting the choroidal point and the middle cerebral artery visualized through the pia of the anterior sylvian fissure. It is important to stay below this line to avoid injury to the globus pallidus and the cisternal segment of the anterior choroidal artery (injury to this vessel is a significant source of hemiplegia and hemianopsia after temporal lobectomy).

The amygdala removal begins with identification of the choroidal point. This is located at the anterior extent of the choroidal plexus where the anterior choroidal artery enters the temporal horn of the lateral ventricle. Once identified, this demarcates the posterior-superior point of resection of the amygdala, as mentioned previously. The surgeon extends an imaginary line across the gray matter of the amygdala from the choroidal point to the middle cerebral artery visualized through the pia of the anterior sylvian fissure. In my experience, this is often difficult to visualize, and the line is extended horizontally from the roof of the temporal horn so that resection in a plane above the roof does not occur. Remember that the goal is to avoid resecting the superior amygdala, which blends imperceptibly into the globus pallidus, and to avoid exposure of the anterior choroidal artery in the cistern. After the superior line of resection is begun, it is carried down to the pia overlying the brainstem, third nerve, and interpeduncular fossa. The remaining dissection should be subpial and will remove temporal polar tissue, lower portion of the amygdala, and uncus. The posterior limit of this stage involves subpial aspiration of the uncus where it joins the head of the hippocampus. During division and removal of this tissue, the free edge of the hippocampal sulcus becomes visible as it arises from the perimesencephalic cistern. When this stage is complete, the anterior free edge of the tentorium, the third nerve, and the anterior aspect of the posterior cerebral artery should be seen.

The hippocampal removal begins with gentle aspiration of the parahippocampal tissue just deep to the remnant of the collateral sulcus. This tissue is removed in subpial fashion and can be carried as far medially as the edge of the tentorium and posteriorly curving deep below the hippocampus to the region of the hippocampal tail. This allows for gentle retraction of the hippocampus down toward the floor of the middle cranial fossa and reduces the need for retraction on the roof of the ventricle as the choroidal fissure is explored. After the parahippocampal tissue is removed, the lateral ventricle sulcus in the posterior aspect of the ventricle is further divided to allow easier access to the tail region of the hippocampus. This step effectively disconnects the overlying temporal cortex from the hippocampus and allows safer removal of the posterior aspect of the hippocampus. During the hippocampal removal, retraction on the roof of the ventricle and on the remaining posterior temporal lobe is undesirable and should be minimized by using the steps described earlier and by changing the position of the microscope to enhance the view of the tissues. Excessive retraction on the roof of the ventricle can lead to a retraction hemiparesis or injury to the optic tract, and similar retraction on the dominant posterior temporal lobe can lead to a postoperative language deficit. After the parahippocampus is removed, the choroidal fissure is exposed by gently retracting the choroid plexus superiorly with the use of a cottonoid patty or Telfa sponge cut to size. The fornix of the hippocampus and dentate gyrus are identified and gently aspirated, which exposes the underlying hippocampal sulcus. This is an extremely important landmark because it contains the arterial and venous supply to the hippocampus. It is much more robust in the head and anterior body region of the hippocampus and becomes more inconsistent in the posterior hippocampal regions. It should be exposed by aspiration of the fornix, dentate gyrus, and uncal tissue so that the entire width is visualized. The anterior extent is a free edge around which the parahippocampus curls to become the uncus. As this tissue is aspirated, the free edge is apparent. The deep origin of the sulcus is the pia overlying the brainstem, and care is necessary to avoid injury. The hippocampal arteries arise directly from the lateral aspect of the posterior cerebral artery, and direct visualization of the course of the hippocampal arteries should be obtained before coagulating and dividing them. This reduces the chances of ligating a traversing vessel that arises from the posterior cerebral artery and crosses the hippocampal sulcus before reentering the perimesencephalic cistern to irrigate the posterior thalamus and brainstem. To reiterate, only those vessels terminating in the hippocampus should be divided. After the arteries are coagulated, divide them so as to leave an arterial tail that is easily reached and coagulated. This avoids a scenario in which the divided stump retracts into the perimesencephalic cistern and causes subarachnoid hemorrhage. After the hippocampal sulcus is divided, the hippocampus and any remaining parahippocampus can be gently peeled off the underlying pia and removed to be sent to pathology for study. The mesial resection is then assessed and any remaining accessible posterior hippocampus removed. The posterior cut across the tail of the hippocampus ideally should allow for 3 to 4 cm of hippocampus to be removed.

When the resection is complete, hemostasis is achieved by time, irrigation, and the judicious use of hemostatic materials. Bipolar coagulation of bleeding points in the mesial temporal or sylvian fissure region is best avoided to prevent injury to underlying structures. Similar bleeding points in the residual amygdala tissue are also best treated with application of hemostatic agents, irrigation, and time. The cavity is then filled in with saline and the dura closed in watertight fashion. The craniotomy flap is attached with titanium fixation and the muscle sewn together and reapproximated to the residual cuff. The skin is then closed in anatomic layers over a subgaleal drain to reduce postoperative swelling.