36 Surgical Failure and Reoperation

10.1055/b-0034-84148

36 Surgical Failure and Reoperation

Blount, Jeffrey P., Kim, Hyunmi, Rozzelle, Curtis J., Kankirawatana, Pongkiat

In children, localization-related medically refractile epilepsy (LRMRE) is inherently neocortical in origin. By far the most common structural anomaly encountered in pediatric surgical epilepsy specimens is malformations of cortical development (MCDs) which are also known as cortical dysplasias (CDs).1 Although complex, they can be suitably and summarily considered structural aberrations of normal neocortex.2,3 The human cerebral neocortex is remarkable in the breadth and extent of its projections and axonal arborizations. Such an inherently robust projection pattern of the neocortex makes initial localization difficult and increases the likelihood of failure of long-term seizure control with focal neo-cortical resections. As a result, epilepsy surgery in children has traditionally been a time and resource consumptive undertaking that has only enjoyed a fraction of the success of adult epilepsy surgery (in which temporal resection is the most common intervention). Surgical series in children have traditionally demonstrated rates of failure of 40 to 50%. Technologic and conceptual advances over the past two decades have resulted in better outcomes but surgical failure, and reoperation, has long been an inherent issue within pediatric epilepsy surgery.4–6 Reoperation can be highly effective but has inherent risks that are often higher than the initial operation.7,8

Surgical Failure Defining Surgical Failure

Although different neurosurgical operations for epilepsy have different objectives, the overarching principle is the elimination of seizures and the attainment of a seizure-free outcome. Clearly the objective of palliative procedures such as corpus callosotomy is more modest and usually includes an elimination of drop events and decrease in overall generalized seizures. Surgical failure is the lack of attainment of the specific objective for a given procedure and usually implies the return of seizures after an operation designed to eliminate or reduce them. The rate of surgical failure varies between procedures and the timeframe of the operative intervention. For example, grid-based focal cortical resections for CD now show an approximately 70 to 75% incidence of seizure-free outcome and a 90% likelihood of marked seizure reduction, whereas large series from the 1970s showed seizure-free outcomes in the 50 to 60% range.1,4,9–12 Hemispherectomy has always been a highly effective procedure with seizure-free outcomes of 80 to 85%.1,13,14 Temporal resection has similarly been highly effective, with 80 to 85% seizure-free rates for adults with mesial temporal sclerosis who undergo anterior temporal lobectomy. Temporal lobe surgery in children is somewhat different in that neocortical and lesional epilepsy predominates, but outcomes in general are similarly good.1,15,16

The traditional definition of surgical failure as a lack of seizure-free outcome must be carefully considered in pediatric epilepsy surgery. Although it is unequivocal that the best outcomes arise from seizure freedom, many of the pediatric epilepsy syndromes are so severe that outcomes other than absolute seizure freedom can be associated with markedly improved quality of life for patients and families. As such, these procedures may fail by conventional analyses that emphasize degree of seizure freedom, yet they are highly successful in substantially reducing the burden of clusters and flurries of daily seizures. It is inaccurate and inappropriate to consider a procedure a surgical failure that reduces a child’s seizure burden from hundreds of events per day to small numbers of events per month for the singular reason that the child is not rendered seizure free. Seizure freedom will always represent the gold standard for epilepsy surgery (and must remain the standard toward which operative intervention strives to achieve). However, for some severe pediatric epilepsy syndromes, it is an artificial gold standard that underestimates the profound impact that reduction of seizure burden may play.

Recognition of Surgical Failure

Acute postoperative seizures (APOSs) are seizures that occur within the immediate 7 to 14 days after surgery.17–19 Traditionally, these events have been attributed to transient local phenomena that have occurred as a result of surgery, such as brain edema, local hemorrhage, or metabolic disturbances. APOSs often have a particularly devastating impact on the patient and the family. As a result, several reports have attempted to determine the predictive capability of APOSs to predict long-term seizure control. Unfortunately, the majority of these studies incorporate adult patients, and a limited number of purely pediatric series exist.18–21 Mani et al showed that APOSs were an independent predictor of poor surgical outcome in a group of 132 pediatric patients undergoing extratemporal resections for intractable epilepsy.18 Extratemporal resection was more frequently associated with the occurrence of APOSs, and long-term seizure control was 27%, 22%, and 13% at 6, 12, and 24 months postoperatively in the group that had demonstrated APOSs. The incidence of APOSs was found to be significantly higher after extratemporal resection than after hemispherectomy.18

Park and colleagues demonstrated that APOSs predicted a lower postoperative seizure-free rate in a group of 148 children and adolescents.21 Twenty-five percent of the patients had APOSs, and this group demonstrated only a 51% rate of long-term seizure control, which contrasted sharply with the 81% rate of seizure freedom seen in non-APOS group. Risk factors included extratemporal lobe surgery, postoperative fever, postoperative interictal activity, and seizures other than partial seizures.21 By contrast, no difference in long-term seizure control was noted between a group of 17 patients (26%) who had APOSs after frontal resection and a larger group (48 patients or 74%) who did not have APOSs in a group of 65 adults reported by Tigaran et al.19 In another study, it was found that more than five seizures within 7 to 10 days of hemispherectomy has been correlated with long-term failure of control and a more complicated hospital course.17

Little other evidence currently exists regarding the timing of surgical failure, although it is increasingly clear that rates of postoperative seizure freedom decline gradually for the first several years after surgery. Thus, it would appear a reasonable response to address the occurrence of postoperative seizures as a significant, but not catastrophic, event. A straightforward, compassionate approach that shares the best available evidence with the family by explaining that postoperative seizures are associated with poorer long-term control yet emphasizes that not all patients having postoperative events demonstrate poor long-term control is the most appropriate response. It is essential that the patient and family understand that the epilepsy team will continue to follow, assess, and treat the patient over time until the epilepsy is controlled or the family and treating team agree that further treatment incorporates unacceptable risk to the patient. When the seizures occur acutely after cortical resection surgery (APOS), it is better to allow the patient time to recover from surgery, re-evaluate seizure frequency and semiology, relocalize if necessary, and then reconsider further medical or surgical therapeutic options. Assurance of commitment to this process on the part of the treating epilepsy team goes a long way toward easing patient and family disappointment, fear, and anxiety and is instrumental in building confidence in the team for the long-term relationship that is often necessary for success in pediatric neocortical epilepsy. Disconnection surgeries constitute an exception to this approach. If the seizures occur initially right after surgery in which disconnection is the primary surgical technique, such as functional hemispherectomy or corpus callosotomy, then prompt consideration for imaging to identify regions of insufficient or failed disconnection would be more appropriate. If a clear, failed area of disconnection can be identified, prompt return to the operating room may prove the easiest and safest route to seizure control or seizure freedom in these patients. Conversely, identifying a singular area of failed resection or disconnection is rarely the case for neocortical grid-based resections.

Prediction of Surgical Failure

No consistent and reliable clinical, electrophysiological, or radiological preoperative indicator of surgical failure has been reported. Cohen-Gadol et al found that normal pathological findings, male gender, prior surgery, and an extratemporal origin for seizures were factors associated with poor outcomes.22 When substratified for CD-induced epilepsy, the same authors found that only a clear and complete resection of all the regions of CD was predictive of good outcome.22 Kim and colleagues defined the extent of resection of CD as an important correlate of surgical outcome.4 The extent of resection of the CD was also found to be the only feature correlating with surgical outcome in pediatric CD in a study published by Krsek and colleagues.23 These investigators reviewed a large cohort of 149 patients with histologically confirmed MCD and specifically surveyed whether age, duration of epilepsy, seizure characteristics (infantile spasms, status epilepticus, secondarily generalized tonic-clonic seizures (SGTCS), neurological or neuropsychological findings, EEG, or MRI characteristics predicted surgical outcome. The only factor found in this series (which is the largest such series to date) was the extent of resection of the region of CD.23 The extent of resection of a CD is conceptually and intellectually important, yet its practical implication is substantially limited because the extent of many CDs is not discernable either by magnetic resonance imaging (MRI) or under the operating microscope. Clearly, for those situations in which the limits of the CD may be defined, it is optimal for seizure control that the entire region of dysplasia be resected. Of course, individual circumstances, such as the relationship of the dysplasia to eloquent cortex, will determine whether all of the dysplastic cortex can be safely removed.

The concept of completeness of resection or disconnection is also a central theme and predictor of failure in other operations as well. The principle cause of surgical failure for corpus callosotomy and functional hemispherectomy is incomplete disconnection.

Patterns of Surgical Failure

Patterns of surgical failure can be characterized according to localization and extent of surgical resection. Preoperative localization is either correct and sufficient or incorrect. Similarly resection may be sufficient or insufficient to achieve seizure freedom.

Localization Correct—Resection Insufficient

Situations in which localization is correct yet seizure-free (or markedly improved) outcomes are not attained often involve cases in which epileptogenic tissue overlaps eloquent cortex. In such a scenario, the resection must be limited to avoid irreversible, severe neurological injury. Such decisions are not always straightforward and must be made carefully; the surgically induced deficit should be less severe than predicted or the fixed deficit induced should have less adverse impact on the child’s overall quality of life than that imparted by the seizure disorder. This is particularly the case for motor deficits and markedly less so for crucial and delicate functions such as memory, speech and language, and frontal disinhibition. Neurological deficits related to these functions can be profoundly serious and are less amenable to postoperative intensive therapy to obtain meaningful improvement in the child’s capability. Often a fixed motor deficit can be readily accommodated (particularly if mild to moderate), whereas insults to memory, speech, and frontal executive function are far more disabling. As such, when the issue of proximity to eloquent tissue arises as a cause for surgical failure, the surgeon and epileptologist must consider whether additional technological adjuncts such as awake techniques with real-time evaluation of function, improved functional imaging on frameless navigation platforms, or surgical techniques such as multiple subpial transections or intragyral resections may play a role to extend the resection safely and meaningfully.24

Another scenario in which surgical failure can follow correct localization is that of technical failure in surgical resection. If the surgical resection is improperly or incompletely performed, it is possible to leave tissue in place that has been localized and implicated as epileptogenic in character. Such an example would be the incomplete resection of an MRI-evident lesion by leaving a portion of the lesion in place. At times, the lesional tissue can blend imperceptibly with surrounding brain, and determination of exact extent of the lesion may be very difficult intraoperatively. Frameless navigation may be useful in preventing such problems, yet it is not an absolute protection because tissue shifts can occur when large lesions are removed that can render initial registration relatively inaccurate. Displacement or mislabeling of subdural grids or misinterpretation or mislabeling of grid maps are uncommon other sources of technical error that can result in surgical failure after correct localization.

Initial Localization and Resection Are Correct and Sufficient, but Evolution of the Epilepsy Induces Surgical Failure

Neocortical epilepsy is notorious in its capability to establish new regions of epileptogenesis and propagation. Although the process of surgical localization is aimed inherently at defining a focal or local region implicated in epileptogenesis, CDs likely involve vast regions of neocortex. After resection, new pathological circuitry can become established that frequently has its epicenter at the margin of the previous resection cavity ( Fig. 36.1 ). The exact mechanism by which CD tissue induces epileptogenesis is incompletely understood, but several mechanisms based on molecular observations of the tissue have been proposed.3 Cytomegalic neurons typify CD tissue and are often found in the most electrophysiologically abnormal regions.25 These abnormal cells appear to have important interactions with inhibitory γ-aminobutyric acid (GABA) pathways (which appear upregulated) and play a central role in epileptogenesis. Altered membrane properties have also been observed. Differential expression of N-methyl-D-aspartic acid (NMDA) subunit receptors and persistence of undifferentiated neurons are also regularly encountered in histological evaluation of CD and may play an important contributory role in epileptogenesis.3,25,26 Two different histopathological subtypes have been defined and stratified (IA, IB, and II) that demonstrate clinically important differences such as seizure outcome and neuropsychological deficits.27 Clinical studies for localization implicate the most active region of epileptogenesis but cannot define the cellular limits of the CD. As a result, all resections in which CD is the predominant pathological finding have an inherent risk for reorganization and surgical failure.

Errors in Localization

Localization of medically intractable epilepsy is based on the principle of concordance of semiology, EEG, and noninvasive functional imaging studies in implicating a particular region of cortex in epileptogenesis. Great advances have been made in each of these techniques such that noninvasive localization is a more accurate process than ever before. EEG advances include use of novel montages and the development of high-density EEG arrays. Functional imaging advances are extensive and include the development and widespread application of techniques in magnetoencephalography (MEG), ictalsingle photon emission computed tomography (SPECT) with subtractionictal SPECT scanning coregistered to MRI imaging (SISCOM), and positron emission tomography (PET). Each of these is discussed elsewhere in this book. Despite these considerable advances localization remains a challenging and imperfect process with a real potential for misinterpretation and errors. Errors in localization account for important sources of surgical failure and need for reoperation.

Fig. 36.1 Surgical failure because of reorganization from malformations of cortical development (MCD)-induced epilepsy. This 14-month-old boy initially presented with severe localization-related medically refractile epilepsy and underwent a grid-based resection. He remained seizure free for 7 months postoperatively, and then seizures recurred. This figure illustrates a magnetic source imaging image featuring dipoles showing the recurrence of medically intractable epilepsy at the resection margin of his frontal MCD. Reoperation was performed and localization confirmed with intracranial electrodes. He has been seizure free for more than 5 years.

Patterns in errors of localization can be recognized and organized. The first type of error occurs when an active area of epileptogenesis fails to be recognized. A common example of this error occurs with dual pathology in which a temporal lobe focus of activity is sufficiently active that a simultaneous frontal region of activity is either not detected or ignored. A more common error occurs when an area that is detected by a test is misinterpreted as being less active or less contributory to the overall epilepsy than is the case. This may occur when disease is multifocal or when semiology or other characteristics of the epilepsy appear to implicate other brain regions. Even though concordance is a useful unifying principle of epilepsy localization, it remains a practical reality that information from functional imaging studies often shows significant discordance and that interpretation (which can be very challenging) is always necessary.

Experienced centers usually quickly get a sense for how concordant the localizing information is for a given patient and develop a degree of confidence for the need for intracranial electrodes. A sincere and earnest desire to help a child with severe, progressive, disabling epilepsy can lead to the ill-advised implantation of electrodes despite modest concordance of noninvasive imaging studies. Such “fishing trips” are characterized by a wide and extensive coverage with electrodes and very modest surgical success rates.

Only gold members can continue reading. Log In or Register to continue