Fu n c t i o n a l hemispherotomy has been used since the. Hemispherotomy: efficacy and analysis of seizure recurrence.

Transcription

1 J Neurosurg Pediatrics 4: , 4: , 2009 Hemispherotomy: efficacy and analysis of seizure recurrence Clinical article Dav i d D. Li m b r i c k Jr., M.D., Ph.D., 1 Pr i t h v i Na r aya n, M.D., 3 Al e x a n d e r K. Po w e r s, M.D., 1 Jef f r ey G. Oj e m a n n, M.D., 4 Ta e Su n g Pa r k, M.D., 1 Mar y Be r t r a n d, M.D., 2 a n d Ma t t h e w D. Sm y t h, M.D. 1 Departments of 1 Neurosurgery and 2 Neurology, St. Louis Children s Hospital, Washington University School of Medicine, St. Louis, Missouri; 3 Princeton Brain and Spine Care, Princeton, New Jersey; and 4 Department of Neurological Surgery, Children s Hospital and Regional Medical Center, University of Washington, Seattle, Washington Object. Hemispherotomy generally is performed in hemiparetic patients with severe, intractable epilepsy arising from one cerebral hemisphere. In this study, the authors evaluate the efficacy of hemispherotomy and present an analysis of the factors influencing seizure recurrence following the operation. Methods. The authors performed a retrospective review of 49 patients (ages years) who underwent functional hemispherotomy at their institution. The first 14 cases were traditional functional hemispherotomies, and included temporal lobectomy, while the latter 35 were performed using a modified periinsular technique that the authors adopted in Results. Thirty-eight of the 49 patients (77.6%) were seizure free at the termination of the study (mean follow-up 28.6 months). Of the 11 patients who were not seizure free, all had significant improvement in seizure frequency, with 6 patients (12.2%) achieving Engel Class II outcome and 5 patients (10.2%) achieving Engel Class III. There were no cases of Engel Class IV outcome. The effect of hemispherotomy was durable over time with no significant change in Engel class over the postoperative follow-up period. There was no statistical difference in outcome between surgery types. Analysis of factors contributing to seizure recurrence after hemispherotomy revealed no statistically significant predictors of treatment failure, although bilateral electrographic abnormalities on the preoperative electroencephalogram demonstrated a trend toward a worse outcome. Conclusions. In the present study, hemispherotomy resulted in freedom from seizures in nearly 78% of patients; worthwhile improvement was demonstrated in all patients. The seizure reduction observed after hemispherotomy was durable over time, with only rare late failure. Bilateral electrographic abnormalities may be predictive of posthemispherotomy recurrent seizures. (DOI: / PEDS0942) Ke y Wo r d s hemispherotomy hemispherectomy seizure epilepsy epilepsy surgery insula disconnection pediatric neurosurgery Abbreviations used in this paper: CUSA = Cavitron ultrasonic surgical aspirator; EEG = electroencephalogram; FH = functional hemispherotomy; MCD = malformation of cortical development; PIH = periinsular hemispherotomy; SWS = Sturge-Weber syndrome. J Neurosurg: Pediatrics / Volume 4 / October 2009 Fu n c t i o n a l hemispherotomy has been used since the 1980s to treat disabling, medically refractory epilepsy resulting from diffuse unilateral hemispheric disease. 12 Over the years, the procedure has evolved as technical variations have been described. 4,11,13,14,16,19,20 Regardless of the specific technique, the operation generally includes a transventricular callosotomy, disruption of the internal capsule/corona radiata, resection of mesial temporal structures, and a frontobasal disconnection. 2,11 Multiple studies have shown FH to be of significant benefit in these patients, with long-term rates of postoperative seizure freedom variously reported in the range of 43 90%. 3 This article contains some figures that are displayed in color on line but in black and white in the print edition. 323

2 D. D. Limbrick et al. Fig. 1. Schematic representations of the 2 hemispherotomy techniques used in the current study. Left: A surface rendering showing the area resected in FH. Below are coronal and axial views showing the areas transected for the disconnection. Note that the anterior temporal lobe has been resected and the insular cortex remains intact (although this was resected in 4 of 14 cases). Right: The surface rendering the limited area of frontal and temporal opercula removed to permit the hemispheric disconnection in PIH. Coronal and axial views illustrate the disconnection. For the PIH, the corpus callosotomy is performed by subpial aspiration following the anterior cerebral arteries. A selective amygdalohippocampectomy is performed rather than a temporal lobectomy. Relatively little data exist regarding seizure recurrence after FH. Perhaps the most important factor in preventing seizure recurrence after FH is proper patient selection. 9 However, even in appropriately selected patients, seizures may recur. Specific pathological conditions (such as hemimegalencephaly or MCDs), 9,10 incomplete disconnection, 8 and residual insular cortex 1 have been implicated in seizure recurrence after FH, but more data are needed to accurately predict outcome following FH and thus select appropriate candidates for this operation. In the present study, we describe our series of 49 patients who underwent hemispherotomy at St. Louis Children s Hospital. Freedom from seizure was achieved in nearly 78% of these patients, and nearly 90% achieved Engel Class I II outcomes. Complications were infrequent and easily managed. In patients who experienced continued or recurrent seizures postoperatively, efforts were made to identify contributing factors and predictors of seizure recurrence. Methods Patient Characteristics Approval by the St. Louis Children s Hospital/Washington University in St. Louis Human Research Program was obtained prior to the initiation of this study; thus, all data were collected in accordance with the standards set forth by the 1964 Declaration of Helsinki. Informed consent was obtained or waived in accordance with the approved Human Research Program protocol (WU HRPO# ). Medical records for all patients undergoing hemispherotomy at St. Louis Children s Hospital were reviewed. Forty-nine patients underwent the procedure between the years of 1995 and The information acquired for each case included patient demographics, clinical history, preoperative studies (MR images, PET and SPECT scans, EEGs, video EEGs, and neuropsychological evaluations), seizure type and frequency, prior neurosurgical procedures, surgical technique for hemispherotomy, operative statistics (operative time and estimated blood loss), histopathology, surgical complications, need for shunt, postoperative seizures, length of hospital stay, duration of follow-up, current antiepileptic drug regimen, and postoperative neurological condition (hemiparesis and cognitive ability). As this is a retrospective study, no prospective criteria were set to assess candidacy for hemispherotomy. Rather, patients were each assessed individually based on their clinical picture (existing hemiparesis, visual field cut, language dominance, neuropsychological testing, and seizure semiology), EEG or video EEG findings, diagnostic test results (MR images, SPECT and PET scans, and the Wada test), and one or more detailed discussions at a multidisciplinary epilepsy surgery conference. Although typical surgical candidates demonstrated severe hemispheric dysfunction with ipsilateral seizure onset only, hemispherotomy was also considered in patients with bilateral epileptiform activity when the majority of seizures emanated from a single hemisphere and the seizures were life-threatening or causing rapid neurological decline. Surgical Techniques From 1995 to 2003, 14 FHs were performed by 2 surgeons at our institution. The surgical technique used in this group was based on the description by Schramm et al. 13 (Fig. 1A). Briefly, the operations began with a temporal lobectomy, which provided access to the temporal horn of the lateral ventricle. A transcortical, perisylvian window was then used to expose the remainder of the lateral ventricle, and through this window a parasagittal transventric- 324 J Neurosurg: Pediatrics / Volume 4 / October 2009

3 Hemispherotomy: efficacy and analysis of seizure recurrence Fig. 2. Pre- and postoperative MR images obtained in a 7-year-old boy who underwent a modified PIH on the left side. A C: Preoperative axial, coronal, and left parasagittal contrast-enhanced T1-weighted images demonstrating left cerebral atrophy with ex vacuo dilation of the left lateral ventricle and prominent subarachnoid spaces. D F: Postoperative MPRAGE T1-weighted images. The axial image (D) shows resection of the frontal and temporal opercula with disconnection to the midline. Solid arrow indicates the mesial occipital disconnection and provides reference for the intraoperative photograph in Fig. 4D. Coronal view (E) demonstrates the transventricular corpus callosotomy (arrowhead). Arrowhead is also seen in intraoperative photograph Fig. 4B. The left parasagittal view (F) shows the frontobasal disconnection, which is performed by subpial aspiration along the proximal anterior cerebral artery. The asterisk in this panel corresponds to the intraoperative photograph in Fig. 5A. ular corpus callosotomy was performed. The frontobasal and mesial occipital areas were then disconnected. The insula was decorticated in 4/14 cases, depending in most cases on the underlying pathological entity (MCDs were decorticated). In 2003, a third surgeon began using a variation of the PIH technique described by Shimizu and Maehara 16 (Fig. 1B). Preoperative and postoperative MR images from a typical case are shown in Fig. 2. The PIH method involves opening the sylvian fissure widely and exposing the insula (Fig. 3A). The extent of the circular sulcus is then defined and dissected, and the frontal and then the temporal opercula are resected and sent for pathological evaluation (Fig. 3B). The frontal horn of the lateral ventricle is entered using Stealth image guidance (Medtronic, Inc; Fig. 3C), and the ventricular opening is continued posteriorly to the atrium of the lateral ventricle and followed to the anterior extent of the temporal horn. Once the temporal horn has been opened, a selective amygdalohippocampectomy is performed (Fig. 3D). The tail of the hippocampus is resected posteriorly and medially, providing the mesial occipital disconnection. From the frontal horn of the lateral ventricle, image guidance and/or Doppler ultrasonography is used to localize the pericallosal arteries through the corpus callosum (Fig. 4A). A transventricular corpus callosotomy is then created and extended from the frontal horn to the mesial occipital disconnection (Fig. 4C and D). A frontobasal disconnection is performed using the anterior cerebral arteries to follow the anterior aspect of the corpus callosotomy from the frontal horn of the lateral ventricle through the basal J Neurosurg: Pediatrics / Volume 4 / October 2009 frontal cortex (Fig. 5A). Finally, the insula is decorticated, leaving the white matter and underlying basal ganglia intact (Fig. 5B). Image guidance is used in all PIH cases, and an external ventricular drain is routinely left in place to clear blood and degradation products. We have performed 35 PIHs using this technique. Outcome Measures and Data Evaluation The primary outcome measure evaluated in the present study was seizure control, reported using the modified Engel classification. 7 Engel Classes I (free of disabling seizures) and II (rare disabling seizures), and Classes III (worthwhile improvement) and IV (no worthwhile improvement) were combined to optimize the power of the statistical analyses (see Engel et al., for the complete classification scheme). Additional variables included recovery from hemiparesis, development and cognitive ability (as assessed by the treating neurologist), and anticonvulsant medication regimen. Surgery type (standard FH versus PIH), preoperative diagnostic testing (MR imaging and EEG), postoperative MR imaging (to assess completion of disconnection), histopathological diagnosis, and residual insular cortex were evaluated for their relative impact on postoperative seizure outcome in an attempt to identify predictors of poor outcome after FH. Variables including seizure outcome, estimated operative blood loss, operative time, and length of hospital stay were used to compare the 2 surgical techniques for hemispherotomy. The mean values ± SDs were used to express each data 325

4 D. D. Limbrick et al. Fig. 3. Intraoperative photographs of PIH obtained in the same boy as in Fig. 2. A: The sylvian fissure is opened widely to the level of the insula with dissection of the circular sulcus underlying the frontal and temporal opercula. B: The frontal and temporal opercula are resected, exposing the insula with overlying middle cerebral artery branches. C: Stealth neuronavigation is used to choose the most appropriate entry point into the ventricle. The ventricle is shown partially opened. The remainder of the ventricle is subsequently opened following the C-shape of the ventricle from the anterior limit of the frontal horn to the anterior extent of the temporal horn. D: The operating microscope is used to perform a selective amygdalohippocampectomy. In this photo, the amygdala has been aspirated with a CUSA, and blunt dissection with a dissector is used to remove the hippocampus. set. Data were analyzed using the unpaired t-test, Wilcoxon test, Fisher exact test, or chi-square test as appropriate, with p 0.05 considered significant. Kaplan-Meier analysis was used to assess seizure outcome over time. The Cochran- Armitage test was used to determine whether there was a linear trend for the proportion of patients in each Engel class across levels of follow-up. The binomial dependent variable was the dichotomized Engel Class (I II or III IV) and the ordered independent variable was follow-up month category (6, 12, 24, > 24 months). Statistical evaluation of the data was performed in conjunction with the Washington University Research Design and Biostatistics Group using SPSS 15.0 (SSPS Inc.) on a Windows XP platform. Results Patient Characteristics Forty-nine patients underwent hemispherotomy performed by 3 surgeons at St. Louis Children s Hospital between 1995 and Fourteen of these patients underwent standard FHs, while 35 received PIHs. Patient age at the time of surgery ranged from 0.2 to 20.5 years. Thirtyone patients (63.3%) were male, and 18 (36.7%) were female. Nineteen operations (38.8%) were on the left side. The mean follow-up period for the study was 28.6 months (range months). Seizure Types Twenty-three patients (46.9%) demonstrated > 1 clinical seizure type. The seizure types included partial (20 patients [40.8%]), complex partial (19 [38.8%]), infantile spasms (10 [20.4%]), generalized tonic-clonic (10 [20.4%]), tonic (2 [4.1%]), epilepsia partialis continua (1 [2.0%]), and subclinical seizures in a minimally responsive patient (1 [2.0%]). Ten patients (20.4%) experienced status epilepticus, and in 5 cases (10.2%) this was acutely life-threatening. Electrographic correlates to these clinical events are discussed below. Seizure Etiology Of the 49 patients who underwent hemispherotomy, the cause of seizure included infarct or encephalomalacia in 13 (26.5%), hemimegalencephaly in 9 (18.4%), cortical dysplasia in 7 (14.3%), SWS in 4 (8.2%), Rasmussen encephalitis in 4 (8.2%), and posttraumatic (including nonaccidental trauma) in 4 patients (8.2%; Fig. 6). Other less common causes included developmental abnormalities: polymicrogyria in 2 patients, and porencephalic cyst, middle cranial fossa arachnoid cyst with associated hypoplastic temporal lobe, recurrent ganglioglioma with ischemic and gliotic change, and intraventricular hemorrhage in 1 patient each. One patient had previously undergone a right frontal topectomy for a radiographically occult ictal onset zone 326 J Neurosurg: Pediatrics / Volume 4 / October 2009

5 Hemispherotomy: efficacy and analysis of seizure recurrence Fig. 4. Intraoperative photographs from the same child as shown in Figs. 2 and 3. A: Doppler ultrasonography is used to locate the pericallosal artery just above the interface between the septum pellucidum (SP) and the corpus callosum (CC). B: Once the artery is located, the CUSA is used to start the corpus callosotomy via subpial aspiration down to the level of the vessel. Arrowhead corresponds to the arrowhead in panel 2E and is provided for orientation to the MR image. C: The transventricular corpus callosotomy continues posteriorly, following the course of the pericallosal artery. D: The corpus callosotomy is continued to the area of the previous hippocampal tail resection to complete the mesial occipital disconnection. Arrow corresponds to the arrow in panel 2D and is provided for orientation to this area on the MR image. localized with electrocorticography. The histopathological report in this case showed only gliosis with neuronal loss. Postoperative Seizure Outcome Using the modified Engel classification, 7 posthemispherotomy seizure outcome was examined. At the last follow-up (mean 28.6 months), 38 (77.6%) of 49 patients of patients were seizure free (Engel Class I), 6 (12.2%) had rare, disabling seizures (Engel II), and 5 (10.2%) showed worthwhile improvement in their seizures (Engel III; Fig. 7). None of the 49 patients demonstrated Engel IV outcome (no worthwhile improvement). To increase the statistical power of the dataset, we further evaluated postoperative outcome by combining Engel Classes I and II and comparing these values with a combination of Engel Classes III and IV. Of the 49 patients in the study, 44 (89.8%) were classified as having an Engel I II outcome, while 5 (10.2%) had Engel Class III IV outcomes. Thus, nearly 90% of patients had an Engel Class I II outcome, and all patients undergoing hemispherotomy experienced at least worthwhile seizure reduction. The seizure reduction observed after hemispherotomy was durable over time. Forty-four of 49 patients had > 6 months of follow-up. When assessing seizure outcome over time, 91, 92, 92, and 83% of patients enjoyed Engel Class I/II outcomes at 6, 12, 24, and > 24 months, respectively (Fig. 8A). The drop from 92 to 83% in the > 24 month group represented the late decline in the condition J Neurosurg: Pediatrics / Volume 4 / October 2009 of a single patient at 71 months postoperatively. To better elucidate the timing of failure, a Kaplan-Meyer analysis was performed (Fig. 8 lower). Three of the 49 patients (6.1%) were classified as having Engel Class III outcomes in the first postoperative month, 1 at 6 months, and 1 at 71 months, as noted above. Furthermore, there was no trend for the Engel class to change over the course of the postoperative interval (Cochran-Armitage exact trend test = 0.67). Outcome According to Cause of Seizure Of the 49 patients in the study, 44 (89.8%) attained Engel Class I/II outcomes, and 5 (10.2%) had Engel Class III outcomes. Of the 5 with Class III outcomes, 3 had hemimegalencephaly and 2 had cortical dysplasia. In total, 3 (33.3%) of 9 patients with hemimegalencephaly and 2 (28.6%) of the 7 with cortical dysplasia had Engel Class III outcomes. When all types of MCDs were considered together (hemimegalencephaly, cortical dysplasia, and polymicrogyria in this series), 5 (27.8%) of 18 had Engel Class III outcomes. To better define the effect of seizure causes on outcome, specifically, recurrence of any seizures (Engel II IV), we analyzed cases in which the patients had been followed up for longer than 6 months. Of the 44 patients in this category, 14 had MCDs: 9 (64%) had Engel Class I outcomes, and 5 (36%) had Engel Classes II IV. Meanwhile, 23 (77%) of 30 patients without MCDs had Engel 327

6 D. D. Limbrick et al. Secondary Outcome Measures Various reports have suggested that elimination or reduction of seizures may be associated with improved neurodevelopmental outcome. To investigate this relationship with respect to hemispherotomy, we reviewed neurocognitive outcome as judged by the treating epileptologist in 44 patients with follow-up > 6 months. Routine follow-up neuropsychological testing was not performed in these patients because of practical limitations in access of many patients to a pediatric neuropsychologist. Neurocognition was improved (or normal) in 39 patients (88.6%), worse in 2 (4.5%), unchanged in 1 (2.3%), and unknown in 2 (4.5%). The observed improvements may be due to seizure reduction, weaning off of antiepileptic drugs, or both. Although our data are insufficient to resolve this issue, it is worth noting that 12 (27.3%) of 44 patients were able to stop taking antiepileptic drugs entirely, 9 (20.5%) were weaned to a single drug, and 21 (47.7%) continued to take more than 1 agent. Antiepileptic drug data for 2 patients were unknown. Fig. 5. Intraoperative photographs from the same child as shown in Figs A: A CUSA is used to complete the frontobasal disconnection by resecting basal frontal cortex and then continuing subpial aspiration along the proximal anterior cerebral artery. Asterisk corresponds to the asterisk in panel 2F to provide orientation to this area on the MR image. B: The final stage of the operation was insular decortication. Class I and 7 (23%) had Engel II IV outcomes. There was no statistical difference in Engel class when evaluated by the presence of MCD (p = 0.48, Fisher exact test). Similarly, the presence of hemimegalencephaly alone was not a significant predictor of Engel class (p = 0.31). Evaluation by Surgical Technique From 1995 to 2003, 2 surgeons (J.G.O. and T.S.P.) performed 14 standard FHs performed at our institution. For further technical details of this operation, please refer to the Methods section and Fig. 1A. In 2003, a third surgeon (M.D.S.) began using a variation of the PIH technique described by Shimizu and Maehara 16 (Fig. 1B, Figs. 2 5); 35 surgeries were performed using the PIH technique. The mean age at surgery was 5.0 ± 4.9 years among those who underwent FH, and 7.2 ± 5.3 years in those who underwent PIH (p = 0.19). Males composed 71 and 59% of those undergoing FH and PIH, respectively (p = 0.41). As would be expected, follow-up was significantly longer in the FH group (4.5 ± 2.2 years vs 1.6 ± 1.2 years in the PIH group; p = 0.003). In comparing the 2 techniques, PIH was found to have statistically longer operative times (6.7 ± 1.5 vs 5.6 ± 1.8 hours; p = 0.05), which may be due to the additional surgical time required to resect the insular cortex, or perhaps due to differences in styles among the neurosurgeons. The percent of estimated blood volume lost during surgery did not differ between the 2 techniques 23.8 ± 0.17% for FH and 29.6 ± 0.21% for PIH (p = 0.39) and there was no difference in the length of hospital stay between patients who underwent FH and those who received PIH (7.9 ± 3.8 vs 8.7 ± 3.0 days; p = 0.40). Of the patients with followup periods longer than 6 months, there was no difference in the number of patients with Engel Class I II outcomes (FH: 79%, PIH: 93%; p = 0.31). Of the 5 patients in the PIH group with a follow-up period shorter than 6 months, 4 had Engel Class I II outcomes. Factors Affecting Seizure Recurrence After Hemispherotomy Of the 49 patients in the study, 13 had at least 1 seizure in the postoperative period. One patient had a single seizure that was believed to be the result of acute hypocalcemia, and 1 patient had 2 febrile seizures during a viral upper respiratory infection that arose 2 months postoperatively. In the remaining 11 patients with postoperative seizures, no medical or other factor was identified as a potential cause; thus, although the semiology of the seizures changed in some patients postoperatively, these seizures were defined as recurrent. Data for these 11 patients were compared with those obtained in 31 patients who were seizure free, to identify potential predictors of seizure recurrence following hemispherotomy. Five seizure-free patients were excluded from the analysis because their follow-up duration was shorter than 6 months. Although patients with bilateral epileptiform abnormalities on their preoperative EEGs appeared to have a tendency toward postoperative seizure recurrence, this did not prove to be statistically significant when analyzed 328 J Neurosurg: Pediatrics / Volume 4 / October 2009

7 Hemispherotomy: efficacy and analysis of seizure recurrence Fig. 6. Pie graph showing the causes of intractable seizures in the 49 patients who underwent hemispherotomy. Number of patients for each etiology is listed in parentheses. Hemimeg = hemimegalencephaly; Polymicro = polymicrogyria. Fig. 7. Bar graph showing posthemispherotomy seizure outcome by Engel class. The bars depict the percentage out of 49 patients with Engel I II or Engel III IV seizure outcomes after hemispherotomy. No patient had an Engel Class IV outcome. J Neurosurg: Pediatrics / Volume 4 / October 2009 by bilateral ictal activity (p = 0.15; Table 1), bilateral interictal activity (p = 0.27), or either bilateral ictal or interictal activity (p = 0.15). In the literature and in our own experience, MCDs appeared to predispose patients to postoperative seizure recurrence and worse postsurgical outcomes. For example, 3 of 9 patients with hemimegalencephaly and 2 of 7 with cortical dysplasia had Engel Class III outcomes, whereas there were no cases of Engel Class III or IV outcomes in patients without MCDs. When taken together, all forms of MCD observed in this study (including cortical dysplasia, polymicrogyria, and hemimegalencephaly) were not statistically associated with postoperative seizures (p = 0.46; Table 1). When analyzed independently, hemimegalencephaly also was not statistically associated with postoperative seizures (p = 0.31). Other factors reported in the literature to be predictive of postoperative seizure outcome include the presence of residual insular cortex and incomplete disconnection. Our data failed to demonstrate a statistically significant effect of either insular decortication or MR imaging confirmed disconnection on postoperative seizure occurrence (p = 0.41 and p = 0.21, respectively; Table 1). Although this study includes a large series of hemispherotomies, the sample size limited our ability to characterize useful predictors of seizure recurrence such as those examined above. For example, using the data presented above for EEG laterality (bilateral vs ipsilateral ictal or interictal epileptiform activity on preoperative EEG) to estimate the sample size required to demonstrate significance, we find that 84 and 28 cases would be required in the nonseizure and seizure groups, respectively (power = 0.80; β = 0.20; α = 0.05). This is substantially more than our current case numbers of 30 and 10 in each group. Similarly, 161 and 54 would be required in the nonseizure and seizure groups, respectively, to adequately assess the diagnosis of hemimegalencephaly as a predictor of seizure recurrence (currently, 31 and 11 are reported here). Further, evaluating MR imaging confirmed disconnection as a predictor would require 105 and 35 cases in the nonseizure and seizure groups. Clearly, given these sample size estimations, a larger number of patients will be needed to definitively characterize useful predictors of seizure recurrence following hemispherotomy. Operative and Nonoperative Complications Ten patients (20.4%) experienced complications. Nonoperative complications included a wound infection (in 1 patient), transient occulomotor nerve palsy (in 1), and sagittal and contralateral transverse sinus thrombosis that resolved with 3 months of aspirin (in 1 patient). Seven patients had surgical complications. Three of these patients required ventriculoperitoneal shunts within the first postoperative month, including 1 child each with hemimegalencephaly, a perinatal infarct, and SWS with aseptic meningitis. One child received a ventriculoperitoneal shunt after being lost to follow-up for 6 years postoperatively. One child with SWS required a subdural peritoneal shunt and later underwent revision of his hemispherotomy incision because of chronic wound drainage. One additional patient had an existing ventriculoperitoneal shunt that required revision 9 days postoperatively despite external ventricular drainage for 4 days posthemispherotomy. Finally, an epidural fluid collection developed in 1 patient and was 329

8 D. D. Limbrick et al. TABLE 1: Summary of the characteristics of patients with seizure recurrence* Fig. 8. Seizure reduction after hemispherotomy is shown to be durable over time. Upper: Bar graph showing the percentage of patients in Engel Classes I II at 6, 12, 24, and > 24 months after hemispherotomy. The number of patients with data available for each time interval is listed above each bar. The drop in % Engel Class I II outcomes observed between 24 and > 24 months was from a single late failure in a patient 71 months postoperatively. Lower: Kaplan-Meier curve depicting probability of Engel Class I II outcome over time. Hash marks denote censored data for individual patient outcome at the last followup. The outcomes in 3 patients were Engel Class III in the first postoperative month, in 1 patient at 6 months, and in 1 patient at 71 months. There was no trend for Engel class to change over the course of the postoperative period (Cochran-Armitage exact trend test = 0.67). treated surgically with a temporary external drain. In total, complications were encountered in 3 of 14 patients after FH, and 7 of 35 after PIH (p = 0.93). Discussion Functional hemispherotomy is generally performed in hemiparetic patients with severe intractable epilepsy arising from 1 cerebral hemisphere and has been shown to be of benefit in this setting. 1,3 6,9,10,12,15,18 In the present study, we reported on our experience with 49 patients who underwent hemispherotomy at St. Louis Children s Hospital. Seizure freedom was achieved in 77% of patients, and nearly 90% achieved an Engel Class I II outcome. Of note, all patients in this series benefited from surgery in terms of seizure reduction. Moreover, the benefit obtained after hemispherotomy was durable over time, with no significant change in Engel class throughout the follow-up Characteristic Postop Seizures No. of Patients w/o Seizures (%) No. of Patients w/ Seizures (%) p Value EEG ictal 0.15 bilateral 3 (10) 3 (30) ipsilateral 27 (90) 7 (70) EEG interictal 0.27 bilateral 11 (37) 6 (60) ipsilateral 19 (63) 4 (40) EEG ictal or interictal 0.15 bilateral 12 (40) 7 (70) ipsilateral 18 (60) 3 (30) MCDs present 0.46 no 22 (71) 6 (55) yes 9 (29) 5 (45) hemimegalencephaly 0.31 no 28 (90) 8 (73) yes 3 (10) 3 (27) insular decortication 0.41 no 6 (19) 4 (36) yes 25 (81) 7 (64) disconnection confirmed on MRI 0.21 incomplete 2 (8) 2 (29) yes 22 (92) 5 (71) * Two patients with postoperative seizures were excluded from the analysis because of an identifiable medical cause of seizures (hypocalcemia in one, febrile seizures in another). Five seizure-free patients with < 6 months follow-up were also excluded from the analysis. Numbers in individual categories vary in some cases because of availability of data. Probability values were computed using the unpaired t-test or the Fisher exact test. period. Although formalized neurodevelopmental testing was not routinely performed in the patients included in this retrospective study, our results confirm previous assertions that hemispherotomy results in an improved cognitive outcome. Surgical techniques for functional hemispherectomy and hemispherotomy have evolved as variations on the procedure. 3,4,11,13,14,16,19,20 Regardless of the specific technique, the operation generally includes a transventricular callosotomy, disruption of the internal capsule/corona radiata, resection of mesial temporal structures, and frontobasal disconnection. 2,11 At St. Louis Children s Hospital, a variation on a standard FH was used prior to Since 2003, however, we have used a modification of the PIH previously described by Shimizu and Maehara. 16 In addition to the procedure detailed in Shimizu and Maehara s report, we routinely use Stealth neuronavigation to simplify entry into the body of the lateral ventricle, and Stealth and Doppler ultrasonography to identify the pericallosal artery prior to starting the transventricular 330 J Neurosurg: Pediatrics / Volume 4 / October 2009

9 Hemispherotomy: efficacy and analysis of seizure recurrence corpus callosotomy. Finally, we routinely leave an external ventricular drain in place at the end of the case to drain operative debris and reduce the risk of postoperative shunt dependency. The authors of a recent report also suggested that external ventricular drainage may reduce the incidence of postoperative fevers and shorten the duration of hospital stay after hemispherotomy. 17 The authors of relatively few studies have evaluated the factors affecting seizure recurrence after hemispherotomy. 1,8 Specific pathological conditions (such as hemimegalencephaly and MCDs), 9,10 incomplete disconnection, 8 and the presence of residual insular cortex 1 have been implicated in seizure recurrence after FH. Our data did not show any statistically significant relationship of these factors to seizure recurrence, probably because of our relatively small sample size. Bilateral epileptiform activity on preoperative EEGs and a diagnosis of hemimegalencephaly appeared to be suggestive of seizure recurrence, but neither factor proved statistically significant. We were especially vigilant about the completeness of disconnection; postoperative MR images were routinely obtained for confirmation in all patients who underwent hemispherotomy. There were 2 cases of incomplete disconnection prior to the implementation of the periinsular technique in 2003 and 2 cases after this technique was adopted. The patients in the 2 earlier cases had Engel Class III IV, and the latter 2 had Engel Class I II outcomes. With only 4 cases of recognized incomplete disconnection, it is impossible to adequately demonstrate whether this parameter is a predictor of seizure recurrence; however, it is reasonable to assume complete disconnection is critical to the success of the operation. Of note, all 4 of our cases of incomplete disconnection were in the setting of hemimegalencephaly, suggesting that the technical constraints and space considerations encountered during the surgery may have impeded complete disconnection. Although we have presented a relatively large series of hemispherotomies in this study (49 cases), our analysis of preoperative predictors of postoperative seizure recurrence was limited by the sample size. Our power calculation suggested that a larger number of cases would be required to demonstrate significance of such predictors. Thus, a multiinstitutional study or meta-analysis may be helpful in defining and characterizing such predictors and their use in preoperative clinical decision making and/or family counseling regarding hemispherotomy. Currently, the most important factor in preventing seizure recurrence after hemispherotomy remains proper patient selection for the operation. 9 However, even in appropriately selected patients, seizures may recur. Several patients in the present study underwent hemispherotomy because of rapid neurological deterioration or life-threatening seizures. In these situations, the operation was offered in the setting of known or suspected bilateral seizure onset where one hemisphere was the predominant contributor of seizures. Seizure palliation rather than complete resolution was the goal in these cases. Even in these difficult cases, hemispherotomy resulted in at least a worthwhile reduction in seizure frequency and stabilization of neurological decline. The concept of palliative hemispherotomy is actively under investigation by our group. J Neurosurg: Pediatrics / Volume 4 / October 2009 Conclusions In the present series of 49 patients with severe, intractable seizures, hemispherotomy resulted in freedom from seizure (Engel Class I) in 77% of cases, with nearly 90% achieving an Engel Class I II outcome. All patients who underwent hemispherotomy received at least a worthwhile benefit from the operation (Engel Classes I III). Moreover, the benefit obtained after hemispherotomy was durable over time, with no significant change in Engel class throughout the follow-up period. Although our sample size limits our ability to make definitive conclusions, bilateral epileptiform activity on preoperative EEGs and diagnosis of hemimegalencephaly may be predictive of posthemispherotomy seizure recurrence. A larger, multiinstitutional study is warranted to identify and characterize such factors influencing seizure recurrence, as they may be useful in selecting appropriate patients for surgery as well as counseling families and managing their expectations for postoperative seizure control. Disclaimer The authors report no conflict of interest concerning the materials or methods used in this study or the findings specified in this paper. Acknowledgment The authors acknowledge Karen Steger-May of the Washington University Research Design & Biostatistics Group for her assistance with the statistical analysis presented in this manuscript. References 1. Cats EA, Kho KH, Van Nieuwenhuizen O, Van Veelen CW, Gosselaar PH, Van Rijen PC: Seizure freedom after functional hemispherectomy and a possible role for the insular cortex: the Dutch experience. J Neurosurg 107: , De Almeida AN, Marino R Jr, Aguiar PH, Jacobsen Teixeira M: Hemispherectomy: a schematic review of the current techniques. Neurosurg Rev 29:97 102, De Ribaupierre S, Delalande O: Hemispherotomy and other disconnective techniques. Neurosurg Focus 25(3):E14, Delalande O, Bulteau C, Dellatolas G, Fohlen M, Jalin C, Buret V, et al: Vertical parasagittal hemispherotomy: surgical procedures and clinical long-term outcomes in a population of 83 children. Neurosurgery 60 (2 Suppl):ONS19 32, Devlin AM, Cross JH, Harkness W, Chong WK, Harding B, Vargha-Khadem F, et al: Clinical outcomes of hemispherectomy for epilepsy in childhood and adolescence. Brain 126: , Di Rocco C, Battaglia D, Pietrini D, Piastra M, Massimi L: Hemimegalencephaly: clinical implications and surgical treatment. Childs Nerv Syst 22: , Engel J Jr, Van Ness PC, Rasmussen TB, Ojemann LM: Outcome with respect to epileptic seizures, in Engel J Jr (ed): Surgical Treatment of the Epilepsies, ed 2. New York: Raven Press, 1993, pp Gonzalez-Martinez JA, Gupta A, Kotagal P, Lachhwani D, Wyllie E, Luders HO, et al: Hemispherectomy for catastrophic epilepsy in infants. Epilepsia 46: , Jonas R, Nguyen S, Hu B, Asarnow RF, LoPresti C, Curtiss S, et al: Cerebral hemispherectomy: hospital course, seizure, developmental, language, and motor outcomes. Neurology 62: , Kossoff EH, Vining EP, Pillas DJ, Pyzik PL, Avellino AM, 331

10 D. D. Limbrick et al. Carson BS, et al: Hemispherectomy for intractable unihemispheric epilepsy etiology vs outcome. Neurology 61: , Morino M, Shimizu H, Uda T, Naitoh K, Kawahara S, Ishiguro T, et al: Transventricular hemispherotomy for surgical treatment of intractable epilepsy. J Clin Neurosci 14: , Rasmussen T: Hemispherectomy for seizures revisited. Can J Neurol Sci 10:71 78, Schramm J, Behrens E, Entzian W: Hemispherical deafferentation: an alternative to functional hemispherectomy. Neurosurgery 36: , Schramm J, Kral T, Clusmann H: Transsylvian keyhole functional hemispherectomy. Neurosurgery 49: , Shimizu H: Our experience with pediatric epilepsy surgery focusing on corpus callosotomy and hemispherotomy. Epilepsia 46 (1 Suppl):30 31, Shimizu H, Maehara T: Modification of peri-insular hemispherotomy and surgical results. Neurosurgery 47: , Sood S, Asano E, Chugani HT: Role of external ventriculostomy in the management of fever after hemispherectomy. J Neurosurg Pediatr 2: , Villemure JG, Daniel RT: Peri-insular hemispherotomy in paediatric epilepsy. Childs Nerv Syst 22: , Villemure JG, Mascott CR: Peri-insular hemispherotomy: surgical principles and anatomy. Neurosurgery 37: , Villemure JG, Vernet O, Delalande O: Hemispheric disconnection: callosotomy and hemispherotomy. Adv Tech Stand Neurosurg 26:25 78, 2000 Manuscript submitted January 22, Accepted May 22, Address correspondence to: David D. Limbrick Jr., M.D., Ph.D., Department of Neurological Surgery, Washington University in St. Louis, St. Louis Children s Hospital, One Children s Place, St. Louis, Missouri limbrickd@nsurg.wustl.edu. 332 J Neurosurg: Pediatrics / Volume 4 / October 2009