Multilobar Resection and Hemispherectomy in Epilepsy Surgery

Multilobar Resections

The rationale of curative surgical procedures aiming at control of focal drug-resistant epilepsy is total inactivation (by resection or disconnection) of the epileptogenic zone (EZ) (i.e., the cortical region of onset and early spread of an ictal discharge). Considering the dynamic features of an ictal discharge, and therefore its spatial-temporal structure, it is not surprising that in a number of patients the EZ may not be localized within the anatomic limits that mark the boundaries between the cerebral lobes. Moreover, the extension of an anatomic lesion that may correlate with seizures does not necessarily correspond to that of the EZ. Indeed, lesions limited to a single lobe may be associated to a multilobar EZ; in addition, anatomic abnormalities as wide as an entire hemisphere may be observed in patients with an EZ involving only a portion of a single lobe.

In patients with symptomatic focal epilepsy, the main problem consists in understanding the complex topographic and functional relationships between an anatomic (presumed to be epileptogenic) lesion and the EZ, and there is no general agreement concerning the strategy that better addresses this issue. Furthermore, in patients suffering from focal epilepsies with negative magnetic resonance imaging (MRI) (so-called cryptogenic epilepsies), the identification and spatial definition of the EZ must be based only on electrical and clinical findings.

Preoperative Evaluation

Invasive Investigations

When noninvasive investigations alone fail to correctly define the EZ, and/or when the EZ (or the anatomic lesion) are suspected to involve highly functional regions, invasive recording with intracranial electrodes is indicated. In many centers, subdural electrodes (arranged in strips or grids and placed through bur holes or craniotomy) are used. In our center, as well as in other institutions in Europe and North America, stereotactic placement of intracerebral electrodes (stereoelectroencephalography [SEEG]) is preferred as the invasive EEG monitoring technique of choice, and it is required in approximately 30% of cases operated on. For each patient, the strategy of intracerebral electrode placement is tailored according to previously acquired clinical, electrical, and anatomic data. An example of customized SEEG exploration is shown in Fig. 102.1 . The technical details of this methodology have been reported elsewhere.

This method can reveal that in some cases the ictal discharges do not respect the anatomic limits of a single cerebral lobe, therefore allowing the diagnosis of multilobar epilepsy. This is a frequent occurrence in patients investigated by SEEG: for instance, more than 40% of the patients operated on after SEEG in our center received a multilobar resection, as opposed to only 10% of cases evaluated by noninvasive investigations.

Presurgical Planning

At difference with most standard unilobar resections (e.g., anterior temporal lobectomy), planning a multilobar resection may be a challenging task, in particular when the EZ, as frequently occurs in epilepsies originating from a wide cortical area, is close to or overlaps with highly eloquent areas. Modern procedures of neuroimage postprocessing allow integrating several morphologic and functional information in a common three-dimensional anatomic space by image coregistration, thus providing the surgeon with an extremely useful tool for surgical planning. Indeed, recent evidence has suggested that the use of multimodal image coregistration in the presurgical planning may both reduce the need for invasive EEG recording and improve seizure outcome. A number of diagnostic elements may be incorporated in the same multimodal scene ( Fig. 102.2 ), including cortical morphology, lesions, vessels, subcortical fascicles, areas of activation at fMR imaging, intracranial electrodes, and PET scan findings , which are rendered available for planning of both SEEG implantations and surgical resections. The extent of the presumed EZ, of the resection area and their relationships with functionally critical structures are therefore more easily evaluated, and this contributes to minimizing surgical risks. By identifying usable anatomic landmarks, including cortical and vascular anatomy, the surgeon may choose the best surgical approach. Datasets resulting from multimodal image postprocessing and coregistration may be also imported in the neuronavigation equipment for intraoperative guidance.

Anesthetic Considerations

Resective surgery for treatment of focal epilepsy does not require anesthetic approaches substantially differing from those commonly used in other intracranial procedures. In our center, resections are performed under general anesthesia, with the patient positioned according to the site of surgery. When resections are performed in areas at risk for motor impairment, intraoperative neurophysiologic monitoring is performed by cortical and subcortical electrical stimulations. In those cases, intravenous anesthesia with propofol and opioids is usually preferred to optimize recording of muscle motor evoked potentials. Prophylactic antibiotics (1 to 2 g of cephamezine given intravenously) are administered at anesthesia induction. Candidates for surgical treatment of severe partial epilepsy are commonly in optimal physical conditions, and their interictal intracranial pressure (ICP) is not elevated. Careful anesthesiologic management (mostly hyperventilation, diuretics, or barbiturates) can be helpful when the volume of the exposed brain suddenly increases. This transitory increase in ICP may suggest the occurrence of an epileptic seizure, the clinical symptoms of which are masked by anesthesia. Except for this infrequent, transitory, and easily controlled incident, no other major intraoperative complications are usually expected in epileptic patients.

Surgical Technique

As compared with classical approaches for unilobar epilepsies, those adopted for multilobar excisions are tailored according to the site and extension of the region involved in the procedure. The supine position is usually preferred even when surgery is performed on the posterior portions of the hemisphere, with the operating table being tilted and the head rotated to obtain a comfortable access to the surgical field. The prone position is mainly used when surgery involves the mesial aspects of the parietal or occipital lobes. Both skin incision and bone flap are designed to fully expose the region to be removed as well as the surrounding cortical areas. In some instances, when a two-step procedure is likely, the surgeon must take into account the possible subsequent step when fashioning skin and bone flaps. In patients who have been previously evaluated by SEEG, the tracks of electrodes can be easily identified on the bone surface, and they are used as landmarks to plan the extent of the bone flap. While opening the dura mater, if a SEEG has been previously performed, care should be taken to separate the adhesions between the inner aspect of the dura and the cortical surface at the electrode entry points. Once the dura is opened, it is very useful to match the vascular surgical anatomy with the stereoscopic angiograms obtained preoperatively, which provide an exquisite three-dimensional view of the sulcal and gyral pattern of the region. Careful inspection of the exposed cortex enables the surgeon to identify the entry points of intracerebral electrodes, which are used to draw the borders of resection.

Neuronavigation is helpful in optimizing accuracy and safety of the procedure, and it is routinely used in epilepsy surgery. Once a volumetric T1-weighted MRI sequence is obtained preoperatively and entered in the neuronavigational module, different anatomic and functional elements can be incorporated in the plan of navigation, including the volume of the lesion as it appears in other MRI sequences, the electrodes trajectories of a previous SEEG, and the volume of cortical areas activated at functional MRI, as well as those of major subcortical bundles detected at DTI-fiber tracking ( Fig. 102.3 ).

With the aid of the surgical microscope, resection is accomplished according to a few simple rules:

• Vessels crossing the region of resection but tributary of, or draining from, regions outside its borders must be identified on the stereoscopic angiograms, isolated, and left intact;

• Subpial dissection should be preferred ( Fig. 102.4 ), especially in mesial regions and along the main fissures, to protect vascular and extra-axial nervous structures (e.g., those in the prepeduncular cistern during mesial temporal removal);

  

• Suction and bipolar coagulation should be avoided whenever possible, to obtain as many unaltered tissue specimens as possible available for histopathologic evaluation.

Like hemispherectomy, which has progressively evolved into hemispherotomy (described later) to minimize the risks of hydrocephalus, hemosiderosis, and intraoperative blood losses, extended multilobar resections may be replaced with disconnective techniques. In our experience, disconnections have been particularly helpful in procedures involving the posterior quadrant of the hemisphere, when the EZ extends to the temporal, occipital, and parietal lobes. In these cases, pure disconnective techniques can be used, as well as a combination of resection and disconnection. In the latter instance, unlike other authors, we prefer to remove the occipital and parietal lobes, and to disconnect the temporal lobe, which may exert a sort of “push-up” effect if left in place, thus limiting residual brain dislocation into the resulting surgical cavity ( Fig. 102.5 ). Like other disconnective techniques, posterior hemispheric disconnections are not indicated if neoplastic pathologies are suspected. A drawback of disconnective approaches is the limited availability of resected tissue, which may prevent a reliable etiologic diagnosis and restrict the access of basic scientists to pathologic specimens.

In past years, it was a standard practice in our institution to administer a prophylactic loading dose of intravenous phenytoin (10 to 15 mg/kg body weight) at dural closure. We have subsequently abandoned this protocol owing to side effects of phenytoin, which often delayed the postoperative mobilization of the patient, and we have replaced it with the immediate reintroduction of the patient’s habitual therapy via nasogastric tube as soon as he or she is moved to the intensive care unit (ICU) after surgery. In selected cases, especially those receiving resections close to the rolandic region, intramuscular phenobarbital may be added at the end of surgery and continued orally for several days postoperatively.

The anatomofunctional basis of morbidity in multilobar resections does not differ substantially from that of unilobar resections, and it is strictly dependent on the site of surgery. Morbidity may include transient motor impairment, akinesia, and mutacism after removal of the supplementary motor area, transient speech disturbances following resections involving the neocortex of the dominant temporal lobe, and contralateral superior quadrantanopia in anterior temporal lobectomies. The risk of larger contralateral visual field defects is higher following resections in the posterior portion of the hemisphere. When removal is contiguous to the rolandic region, the risk of motor-sensory impairment is usually increased, and functional mapping by either intracerebral electrical stimulations at SEEG, as well as functional MRI or intraoperative cortical and subcortical neurophysiologic monitoring, is helpful to minimize the chances of unwanted postoperative deficits.

Outcome on Seizures

The reported rate of patients achieving seizure freedom after a multilobar resection is highly variable across series, ranging from 24.5% to 92% ( Tables 102.1 and 102.2 ). This variability is probably ascribable to several factors, which include, for instance, different selection criteria of surgical candidates, presurgical diagnostic work-up, and etiologic substrates. Although it has been observed that in children with normal MRI there is no difference in seizure outcome between patients receiving unilobar or multilobar resections, the general opinion is that patients with multilobar epilepsies have less chances to obtain seizure freedom after surgery compared with unilobar cases. Notably, multilobar resections seem to provide better results when performed in the posterior hemispheric quadrant compared with other locations. However, it must be stressed that, although multilobar epilepsy represents a complex epileptologic situation which could hardly be improved by further drug regimens, a substantial number of patients can be cured or improved (Engel classes I to III, approximately 86% in our series) by surgical treatment.

Illustrative Case

This right-handed young girl was 11 years old at surgery. Delivery occurred at term, and it was complicated by umbilical cord wrapping around her neck and by a left clavicle fracture. In the first months of life a left upper limb underutilization became evident. The first seizure occurred at 15 months; 2 years thereafter the child presented a status epilepticus, controlled with valproate. Seizures continued despite several attempts with antiepileptic drugs. A brain MRI disclosed diffuse right hemispheric atrophy, with ipsilateral ventricle enlargement and hyperintense white matter and hippocampus, which were ascribed to a possible perinatal hypossic etiology ( Fig. 102.6A–C ). At referral, the neurologic exam disclosed a mild left hemiparesis, and a visual field evaluation documented a left complete hemianopia. She presented with weekly seizures, characterized by a prolonged sensation of fear, nausea or disgust, no clear loss of contact, inconstant late oral automatisms, and, occasionally, speech impairment. Interictal scalp EEG revealed epileptiform abnormalities over the temporoparietooccipital regions of the right hemisphere (see Fig. 102.6F ), and video-EEG recording showed ictal electrical modifications over the same areas, with distinct contralateral propagation in some instances. ¹⁸ FDG-PET indicted decreased metabolism in the right temporal and parietal regions (see Fig. 102.6D and E ). To better define the area to be resected, a SEEG exploration, covering the right temporal lobe, as well as the ipsilateral occipital and parietal regions, was performed ( Fig. 102.7A–F ). Depth recording revealed that ictal onset occurred in the mesiotemporal structures, with subsequent propagation to the parietal and occipital regions (see Fig. 102.7G ). Basing on SEEG evidence, the patient underwent the surgical disconnection of the right temporal, occipital, and parietal lobes ( Fig. 102.8A–D ). Histologic examination of the small amount of tissue removed for disconnection was positive for gliosis. During the first 6 months after surgery, the patient presented rare left-sided motor seizures, and she was then seizure free for the following 2 years of follow-up (Engel class Ic).