Epilepsy is a common neurological comorbidity in. Epilepsy surgery in patients with autism

Transcription

1 clinical article J Neurosurg Pediatr 19: , 2017 Epilepsy surgery in patients with autism Malgosia A. Kokoszka, PhD, 1 Patricia E. McGoldrick, MSN, MPA, CNP, 2 Maite La Vega-Talbott, MD, 2 Hillary Raynes, MD, 2 Christina A. Palmese, PhD, 2 Steven M. Wolf, MD, 2 Cynthia L. Harden, MD, 2 and Saadi Ghatan, MD 1 Departments of 1 Neurosurgery and 2 Neurology, Mount Sinai Health System, New York, New York Objective The purpose of this study was to report outcomes of epilepsy surgery in 56 consecutive patients with autism spectrum disorder. Methods Medical records of 56 consecutive patients with autism who underwent epilepsy surgery were reviewed with regard to clinical characteristics, surgical management, postoperative seizure control, and behavioral changes. Results Of the 56 patients with autism, 39 were male, 45 were severely autistic, 27 had a history of clinically significant levels of aggression and other disruptive behaviors, and 30 were considered nonverbal at baseline. Etiology of the epilepsy was known in 32 cases, and included structural lesions, medical history, and developmental and genetic factors. Twenty-nine patients underwent resective treatments (in 8 cases combined with palliative procedures), 24 patients had only palliative treatments, and 3 patients had only subdural electroencephalography. Eighteen of the 56 patients had more than one operation. The mean age at surgery was 11 ± 6.5 years (range years). At a mean follow-up of 47 ± 30 months (range months), seizure outcomes included 20 Engel Class I, 12 Engel Class II, 18 Engel Class III, and 3 Engel Class IV cases. The age and follow-up times are stated as the mean ± SD. Three patients were able to discontinue all antiepileptic drugs (AEDs). Aggression and other aberrant behaviors observed in the clinical setting improved in 24 patients. According to caregivers, most patients also experienced some degree of improvement in daily social and cognitive function. Three patients had no functional or behavioral changes associated with seizure reduction, and 2 patients experienced worsening of seizures and behavioral symptoms. Conclusions Epilepsy surgery in patients with autism is feasible, with no indication that the comorbidity of autism should preclude a good outcome. Resective and palliative treatments brought seizure freedom or seizure reduction to the majority of patients, although one-third of the patients in this study required more than one procedure to achieve worthwhile improvement in the long term, and few patients were able to discontinue all AEDs. The number of palliative procedures performed, the need for multiple interventions, and continued use of AEDs highlight the complex etiology of epilepsy in patients with autism spectrum disorder. These considerations underscore the need for continued analysis, review, and reporting of surgical outcomes in patients with autism, which may aid in better identification and management of surgical candidates. The reduction in aberrant behaviors observed in this series suggests that some behaviors previously attributed to autism may be associated with intractable epilepsy, and further highlights the need for systematic evaluation of the relationship between the symptoms of autism and refractory seizures. Key Words ASD; autism spectrum disorder; resective; palliative; corpus callosotomy; VNS; vagus nerve stimulator; neuromodulation; epilepsy Epilepsy is a common neurological comorbidity in children with autism. An estimated 5% 40% of autistic children experience epileptic seizures, and the association tends to be stronger in more severely affected patients. 9,25 The burden of medically refractory epilepsy (MRE) has been well established in the pediatric population, and when combined with autism, the cost to the child, caregivers, and society is inevitably higher. 3 Epilepsy surgery, although still an underutilized treatment in the armamentarium against medically refractory seizures in children 39 and adults, 6,13,22 is effective in controlling seizures and bringing about tangible gains in health-related Abbreviations AED = antiepileptic drug; ASD = autism spectrum disorder; DSM-5 = Diagnostic and Statistical Manual, 5th Edition; EEG = electroencephalographic; MRE = medically refractory epilepsy; MTS = mesial temporal sclerosis; PDD-NOS = pervasive developmental disorder, not otherwise specified; RNS = responsive neurostimulation; TS = tuberous sclerosis; VNS = vagus nerve stimulator. SUBMITTED January 25, ACCEPTED July 7, include when citing Published online November 25, 2016; DOI: / PEDS J Neurosurg Pediatr Volume 19 February 2017 AANS, 2017

2 Autism and epilepsy surgery quality of life and development in lesional as well as nonlesional epilepsy from both temporal and extratemporal sources. However, the few reports on outcomes of epilepsy surgery in children with autism show mixed results, and some suggest concerns over the implications of the presence of autism for a worthwhile surgical outcome. 10,30,40 We describe our experience at a comprehensive epilepsy center with a large population of children with autism, where a highly tailored, multimodality surgical approach was used to mitigate the seizure burden associated with these comorbidities. Methods Study Design Approval for retrospective review of existing medical records with waiver of patient consent was obtained prior to collection and analysis of deidentified data. Pertinent medical records were reviewed for patients with autism who underwent surgical treatment for MRE between January 2005 and May 2014, including physician referrals, neuroimaging, genetic testing, neuropsychological records, surgical notes, pathology findings, and neurology and neurosurgery follow-up notes. Patients with the following diagnoses: autism; Asperger syndrome; Rett syndrome with autistic features; and pervasive developmental disorder, not otherwise specified (PDD-NOS) were included, regardless of etiology. These patients are collectively referred to throughout this report as those with autism spectrum disorder (ASD), based on recent changes in the Diagnostic and Statistical Manual, 5th Edition (DSM-5) guidelines for autism diagnosis. 1,2 Study Participants Fifty-six consecutive surgical patients with ASD were identified (Table 1): 49 patients with autism, 2 boys with Asperger syndrome, 2 girls with Rett syndrome, and 3 children with PDD-NOS. There were no patients with childhood disintegrative disorder (known as CDD). 1 Comorbid genetic disorders included tuberous sclerosis ([TS], 3 patients), Sturge-Weber syndrome (1 patient), Angelman syndrome (1 patient), Costello syndrome (1 patient), Cornelia de Lange syndrome (1 patient), and Down syndrome (1 patient). Additionally, 3 patients had point mutations in genes with a known link to epilepsy (SCN1A, PCDH19, and PRICKLE1, all revealed after surgery), and 3 more had at least 1 family member with a history of seizures. Ten patients had some form of migrational abnormality or abnormal cortical organization, 2 patients had brain lesions related to prior medical procedures, 9 had mesial temporal sclerosis (MTS), 3 had epilepsy associated with tumor, and 4 had a presumed vascular etiology for their epilepsy (3 childhood stroke patients were previously published). 19 To our knowledge, 24 of the 56 patients had no comorbidities, structural abnormalities, or genetic mutations associated with an increased risk of autism and epilepsy (Table 2). The following DSM-5 criteria were used to classify autism severity: 1 = mild individual resists changes in routine, requires mild support due to difficulty with effective communication and reciprocal social interactions; TABLE 1. Summary of clinical characteristics of surgical candidates and surgical procedures performed in patients with ASD and epilepsy Patient Characteristics Value Demographics Mean age at surgery (yrs) 11 ± 6.5, range Mean age at 1st surgery (yrs) 10 ± 6, range Sex 39 M, 17 F Epilepsy Mean duration (yrs) 9 ± 5, range 1 28 Mean age at onset (yrs) 1.5 ± 3, range 0 12 Etiology 32 structural &/or genetic, 24 idiopathic Autism Severity 45 severe, 8 moderate, 3 mild Speech 30 nonverbal, 26 verbal Surgical treatments Resective Focal &/or lobar 7 temporal, 5 frontal, 1 HH Multilobar 4 hemispherotomy, 12 subhemispheric Side* 17 lt, 11 rt Palliative CC 26 VNS 14 Invasive monitoring Staged resections 25 Diagnostic only 4 bilat strips, 3 grids & strips CC = corpus callosotomy; HH = hypothalamic hamartoma. Epilepsy duration before initial surgery is given. All procedures performed at our center, including invasive monitoring, were considered when calculating age at surgery. Severity of ASD was graded according to the DSM-5, as described in Methods. * No side was assigned to the resection of the HH; therefore the sides only total = moderate obvious social impairment requiring substantial support in daily functioning, reduced expressive and receptive communication, and elevated levels of distress when interrupted or redirected; and 3 = severe very limited speech and facial or other nonverbal expressions, very limited receptive communication, marked distress when disrupted or redirected from fixation, unable to establish relationships beyond those with caregivers, and unable to function without very substantial support. 2 Individuals 2 years of age or older who use fewer than 5 words purposefully and meaningfully each day were considered nonverbal. 43 Assessment of Outcomes Seizure outcomes were evaluated using the Engel classification system 12 (Fig. 1, Table 2). Additionally, current antiepileptic drugs (AEDs) were compared with preoperative prescriptions. Functional outcomes were assessed based on available neuropsychological records as well as caregivers accounts and direct evaluation during follow-up visits. J Neurosurg Pediatr Volume 19 February

3 M. A. Kokoszka et al. TABLE 2. Detailed clinical characteristics of individual surgical candidates, surgical treatments, and outcomes of epilepsy surgery in patients with ASD Case No. Sex Age at Op (yrs) Severity of ASD Related Diagnoses Radiology Findings Pathology Findings Surgical Procedure(s) Seizure Outcome (FU, mos) 1 M 12 3 nonverbal Hemimegalencephaly Near complete lt hemispheric Neuronal loss &/or gliosis Lt hemispherotomy ID (59) dysplasia 2 M 8, 12 1 Normal Normal CC, staged lt TL IIC (45) 3 M 6 3 nonverbal Lt temporal cortical dysplasia Cortical dysplasia Staged lt TL IA (39) 4 F 18 3 nonverbal Normal NA CC IIIA (36) 5 M 17 3 nonverbal Normal NA VNS IIB (45) 6 M 8 3 Lt parietofrontal polymicrogyria NA CC IIIA (18) 7 M 12, 14 3 nonverbal Normal NA VNS, CC IVC (102) 8 F 7 3 nonverbal FHx Normal NA CC IIIA (11) 9 M 15 1 ALL Rt Fr posttraumatic gliosis NA CC IB (73) 10 M FHx Lt temporoparietal cortical Cortical dysplasia Staged lt TL + PL IA (79) dysplasia 11 M 10 2 Normal Normal Staged lt TL IA (77) 12 M 23, 28 3 nonverbal Normal NA VNS, CC IC (85) 13 F 11, 12 3 SCN1A Lt MTS Normal Staged lt TL, VNS IIIA (102) 14 M 28, 35 3 nonverbal Lt Fr leukoencephalomalacia Polymicrogyria, gliosis, VNS, staged lt FL IIIA (10) white matter heterotopia 15 F 3 3 nonverbal Normal NA CC IIIA (15) 16 F 6 2 Normal NA CC IIA (27) 17 M 8 3 Sturge-Weber Sturge-Weber, rt hemispheric Sturge-Weber, MTS Rt hemispherotomy IA (15) atrophy, rt posterior focal lesion 18 M 12, 14 3 nonverbal Rt occipitotemporal cortical Cortical dysplasia CC, staged rt TL + Fr IIIA (38) dysplasia + Oc resection 19 F nonverbal Normal NA VNS IIA (35) 20 M 10 3 nonverbal Normal NA CC IIIA (14) 21 M 12 3 MCA CVA Rt frontoparietal opercular MTS Staged rt hemispherotomy IA (52) lesion 22 M 13, 17 3 nonverbal Normal Normal CC, staged lt TL + Fr IIIA (35) disconnect 23 F PCDH19 Normal Normal Staged lt cingulate IIIA (35) resection 24 F 17, 19, 20 3 Normal Normal VNS, CC, staged rt FL IIC (80) 25 M 8, 10 3 Normal NA CC, VNS IB (100) 26 M 11 2 AVM Rt temporal porencephalic Ischemic damage w/ Staged rt TL + partial IA (40) cyst, rt occipital atrophy encephalomalcia FL + partial OL 27 M 8 2 Normal NA VNS, CC IA (37) 28 M 5 3 HH HH Resection of HH IA (42) 29 F 19 3 nonverbal Ischemic Lt temporoparietal infarct, bilat MTS Staged lt TL + OF IIA (53) event encephalomalacia resection 30 F 17 3 nonverbal Rett syndrome Normal NA CC IVB (22) 31 F 9 3 nonverbal Normal NA CC IIIA (75) 32 M 4 3 Rt parahippocampal mass Glioma Staged rt TL IA (62) 33 M 8 2 Lt Fr cortical dysplasia Mild cortical dysplasia Staged lt Fr resection IA (91) 34 M 15 3 nonverbal Normal NA Bilat strips, CC IIA (30) 35 F 8, 10 3 Down syndrome Normal NA CC, VNS IIIA (50) 36 F 4 3 nonverbal Normal NA CC IA (2) 37 M 4 3 Normal NA CC IVC (47) CONTINUED ON PAGE 199» 198 J Neurosurg Pediatr Volume 19 February 2017

4 Autism and epilepsy surgery» CONTINUED FROM PAGE 198 TABLE 2. Detailed clinical characteristics of individual surgical candidates, surgical treatments, and outcomes of epilepsy surgery in patients with ASD Case No. Sex Age at Op (yrs) Severity of ASD Related Diagnoses 38 M 16 3 nonverbal Cornelia de Lange syndrome Radiology Findings Pathology Findings Surgical Procedure(s) Bilat MTS MTS Bilat strips, staged lt TL + OF resection 39 M 19 3 nonverbal Normal Elements resembling balloon cells, no true cortical dysplasia Staged lt TL + Fr disconnect Seizure Outcome (FU, mos) IA (47) IIIA (21) 40 F 8 3 nonverbal PRICKLE1 Lt MTS MTS Staged lt TL + OF IIIA (37) resection 41 M 7, 10 3 nonverbal Normal NA VNS, CC IID (105) 42 M 16 3 nonverbal FHx Normal Normal Staged lt TL + Fr IIIA (14) disconnect 43 F 7 3 Rett syndrome Rt MTS MTS, orbitofrontal cortical Staged rt TL + OF IIA (26) dysplasia resection 44 M 8, 14 3 nonverbal Costello Normal NA VNS, CC IIIA (29) syndrome 45 M 7 2 Rt Fr cystic encephalomalacia, MTS Staged rt TL ID (13) lt cerebellar arachnoid cyst 46 M 4, 5, 7 3 nonverbal Normal Normal (Lt partial TL), CC, IIIA (76) staged lt TL + Oc-P disconnect + partial FL 47 M 8, 10, 11, 3 nonverbal TS Multiple tubers NA (VNS, staged rt TL), IIA (100) 12 bilat strips, CC 48 M 5, 7 3 nonverbal TS Multiple tubers, rt temporal Cortical tuber VNS, staged rt TL IIA (51) lobe expansion 49 M 19 3 Normal NA CC IIIA (21) 50 M 2 2 nonverbal Normal Normal Staged rt TL + Fr IA (15) disconnect 51 F 7 3 Angelman Normal NA VNS IIB (117) syndrome 52 M 17 3 Hemophilia, Extensive rt hemispheric MTS Rt hemispherotomy IA (10) CVA encephalomalacia 53 M 2 3 nonverbal TS Multiple tubers Cortical tuber Bilat strips, staged IA (8) lt FL 54 F 3 3 nonverbal Lt mesial temporal atrophy NA Implant; no resection IIA (17) 55 M 9 2 Lt temporooccipital encephalomalacia NA Implant; no resection IA (75) 56 M 6 3 Rt Fr polypachygyria NA Implant; no resection IIA (68) ALL = acute lymphoblastic leukemia; AVM = arteriovenous malformation; CVA = cerebrovascular accident; FHx = family history of seizures; FL = frontal lobectomy; Fr = frontal; FU = follow-up; MCA = middle cerebral artery; NA = not applicable; Oc = occipital; OF = orbitofrontal; OL = occipital lobectomy; P = parietal; PL = parietal lobectomy; TL = temporal lobectomy. Seizure outcomes were measured using the Engel classification. Severity of autism was graded as follows: 1 = mild, 2 = moderate, and 3 = severe, as described in Methods. In the Surgical Procedures column, a plus sign indicates procedures done during the same operation, and commas separate subsequent surgeries; procedures done at another center are in parentheses. Outcome and follow-up of the most recent surgery is given. Disruptive behaviors observed in the clinical setting, such as unprovoked aggression, intentional breaking of hospital property, oppositional and combative behavior during clinical procedures, and attempts to manipulate caregivers by threatening self-injury or violence, were considered aberrant behaviors. Absence of these disruptive behaviors during follow-up visits was considered marked improvement. Nonviolent autistic mannerisms (e.g., rocking, arm waving, odd repetitive sounds) were not included in the aberrant category. Parental reports of changes noted since surgery in the child s overall health, behavior, and cognitive ability were J Neurosurg Pediatr Volume 19 February

5 M. A. Kokoszka et al. FIG. 1. Bar graph showing seizure outcomes of epilepsy surgery in patients with ASD. Palliative procedures performed prior to definitive resective treatments are not shown. In the only case in which a palliative procedure was performed subsequent to resection, outcomes of both surgeries are shown (VNS and temporal lobectomy). The mean follow up was 47 ± 30 months (range months); see text and Table 2 for individual follow-up times. CC = corpus callosotomy. collected from neuropsychological records and follow-up notes. Annual assessments for patients residing in group homes were reviewed along with parental reports. Functional categories listed in Table 3 were based on the most frequently noted changes. When sufficient details were available, patient status in specific areas was interpreted as better, same, or worse than before surgery. The 2-tailed Fisher s exact test was performed to assess statistical significance of changes after surgery; p < 0.05 was considered significant. Any documented comments regarding parents satisfaction with surgery and experience of the overall process were also collected, especially when most difficult part, main benefit [other than seizure control], and overall positive/negative outcome were explicitly stated (Table 4). The chi-square test was used to assess the association of clinical features and Engel outcome. Maximum follow-up was available for 52 of the 56 patients (1 patient died after an accident unrelated to epilepsy 3 years after surgery follow-up as of time of death is given; and 3 patients were lost to follow-up their outcome is given as of the last visit). For patients who had multiple interventions, follow-up times for individual procedures are calculated through the subsequent surgery. The age and follow-up times are reported as the mean ± SD. Results Surgical Procedures and Seizure Control In our population of patients with epilepsy and autism, a total of 69 surgeries were performed at our center, including 25 staged resections and/or disconnections with invasive monitoring, 3 single-stage hemispherotomies, 1 lesionectomy, 26 corpus callosotomies, and 14 vagus nerve stimulator (VNS) placements. In addition, 4 patients had bilateral strip electrodes implanted in a separate procedure, and 3 more patients underwent invasive monitoring with grid and strip electrodes without a secondary resection. One-third of the patients (18 of 56) had more than one operation, and 2 underwent prior epilepsy surgery before referral to our center. Seizure outcomes are summarized in Fig. 1 and detailed in Table 2. Resective outcomes are discussed below with regard to lesional or nonlesional etiology, extent of resection, and whether the temporal lobe was involved, followed by a discussion of palliative procedures. Resections in Patients With Structural Lesions Of the 29 patients who underwent focal or multifocal resections and/or disconnections, in 20 patients their epilepsy was associated with abnormal radiological findings. Three hemispherotomies were performed in patients with extensive hemispheric lesions and baseline hemiparesis, and all 3 resulted in seizure freedom (follow-up 59, 15, and 10 months). Another hemispherotomy was performed in a boy with a small parietal lesion associated with a vascular insult (Case 21), in whom invasive monitoring confirmed diffuse seizure onsets involving the entire hemisphere (Engel IA, follow-up 52 months). 200 J Neurosurg Pediatr Volume 19 February 2017

6 Autism and epilepsy surgery TABLE 3. Parent-assessed functional outcomes of epilepsy surgery Outcome Parents Reports of Changes After Op No Better Change Worse p Value Functional area Areas related to autism diagnosis Speech & language <0.01 development Nonverbal communication <0.001 skills Relating to others <0.001 Adaptation to change >0.05 Cognitive function Attention span <0.05 Memory & learning <0.001 Following directions <0.001 Academic progress <0.01 Aberrant behaviors Overall severity of all symptoms* <0.05 Impulsivity >0.05 Aggression >0.05 Self-injury >0.05 Overall severity of behavioral symptoms by seizure outcome* Engel I >0.05 Engel II >0.05 Engel III >0.05 Engel IV >0.05 Outcomes were not assessed for patients for whom recent follow-up was not available; see Methods for details of data collection and analysis. * The same overall findings are first presented in aggregate and then grouped by seizure outcome. Six patients had MRI evidence supporting a temporal seizure focus (either MTS, cortical dysplasia limited to the temporal region, a temporal epileptogenic tuber, or abnormality in the parahippocampal gyrus). One additional patient (Case 38) had bitemporal MTS, but one-sided seizure onsets were demonstrated with bilateral subdural strip electrodes. Four staged temporal lobectomies were performed in these 7 patients, and 3 staged multilobar resections involved the temporal lobe. Three of these patients remain seizure free since surgery (follow-up 62, 47, and 39 months), and 2 more are mostly seizure free (follow-up 51 and 26 months). The patient in Case 40 (MTS, focal electroencephalographic [EEG] findings) enjoyed 2 years of seizure freedom, after which a reduction in medication dose was followed by new-onset generalized seizures with a bilateral Lennox-Gastaut like EEG pattern never observed prior to surgery. She was recently diagnosed with a mutation in the PRICKLE1 gene. The patient in Case 13 (MTS, focal EEG findings) experienced only 6 months TABLE 4. Long-term benefits outweigh the stress of surgery Parents Perspective: Most Common Parental Comments on Outcome No. of Cases Most important benefit Increased awareness &/or cognitive improvement 13 Patient s happiness & improved quality of life 10 Reduced AED side effects 7 Most difficult part Parents anxiety over outcome, stress during surgery 12 Recovery 11 Decision 9 Overall outcome Positive 43 No improvement 2 Equivocal 4 Benefits and difficulties most frequently mentioned by parents of patients with recent follow-up are listed (see Methods). High parental satisfaction with outcome was noted, with improvements in mood and cognition perceived as the most important benefits of improved seizure control. Comments on difficulties were typically associated with the short-term stress of surgery. of seizure freedom after temporal lobectomy, and subsequently an SCN1A mutation was diagnosed. Four more multilobar resections were performed in cases in which MRI abnormalities involved, but were not limited to, the temporal lobe. Two boys in this group achieved a long-term Engel IA outcome (follow-up 79 and 40 months); 1 girl with vascular etiology achieved an Engel Class II outcome (follow-up 53 months) with newonset generalized seizures following a period of 2.5 years of seizure freedom; and 1 boy (Case 18) achieved an Engel Class III outcome (follow-up 38 months), despite surgical removal or disconnection of epileptogenic areas of cortical dysplasia. The only extratemporal lesional cases were 3 staged frontal lobectomies, 1 of them associated with TS (2 Engel IA, follow-up 91 and 8 months; 1 Engel IIIA, follow-up 10 months). One additional patient with extratemporal lesions underwent a temporal lobectomy, based on subdural recordings pointing to temporal seizure onsets (Engel ID, 13-month follow-up). In the final patient in this group, single-stage removal of a hypothalamic hamartoma resulted in seizure freedom (42-month follow-up). Additionally, 3 patients with abnormal MRI findings whose ictal onsets could not be definitively localized with scalp EEG recordings (Cases 54 56) underwent implantation of subdural electrodes without subsequent resection, because no seizures or only atypical seizures occurred during monitoring. All 3 patients are currently responding well to pharmacotherapy, in 1 case supplemented by experimental use of cannabidiol. Resections in Patients With Nonlesional Epilepsy Nine of the 29 patients in our series who underwent resective procedures were considered nonlesional cases after dedicated epilepsy protocol MRI was performed. Four focal or lobar and 5 multilobar resections and/or disconnections were performed, all aided by invasive monitor- J Neurosurg Pediatr Volume 19 February

7 M. A. Kokoszka et al. ing. The 3 lobectomies resulted in Engel IA (follow-up 77 months) and IIC outcomes (follow-up 80 and 45 months), and resection of the cingulate gyrus resulted in an Engel IIIA outcome (follow-up 35 months). All 5 nonlesional multilobar resections involved the temporal lobe. One resulted in seizure freedom (follow-up 15 months) and 4 resulted in seizure reduction (Engel IIIA, follow-up 76, 35, 21, and 14 months). Neuropathological examination revealed abnormalities resembling balloon cells in the amygdala of the patient in Case 39, but not true cortical dysplasia. Normal tissue structure and cellular morphology were found in the other 4 patients with nonlesional epilepsy who had an Engel IIIA outcome. The patient in Case 23 was later diagnosed with a PCDH19 mutation, following the gene s discovery. Complete Corpus Callosotomy Twelve patients had corpus callosotomy as their only treatment, and 5 more had only callosotomy following prior implantation of a VNS. Six of these 17 patients became seizure free or almost seizure free after callosotomy (mean follow-up 55 months, range months); 8 patients experienced recurrence of seizures at reduced frequency and/or severity (mean follow-up 27 months, range months); 1 had no worthwhile seizure reduction (follow-up 22 months); and 2 reported seizure worsening (follow-up 47 and 102 months). In addition, subsequent laterality was observed in 4 of these 17 patients, 1 of whom (Case 8) has already been deemed a candidate for resection. Similarly, 5 of the resective procedures included in this report followed an earlier corpus callosotomy, which had led to localization of seizure foci. Two more callosotomies were performed after unsuccessful attempts to lateralize seizure onsets with bilateral strip electrodes, in 1 case related to multiple tubers, and resulted in good lasting seizure control (Engel IIA, 100- and 35-month follow-up). Two additional callosotomy cases in which seizures persisted and did not demonstrate focality were followed by implantation of a VNS (see below). Placement of a VNS Fourteen patients had a VNS placed at our center, and in 11 of those the VNS was the initial epilepsy surgery. In 3 of the 11 cases, the VNS provided a significant benefit when combined with medical therapy (Engel II, follow-up 117, 45, and 35 months), and 8 of the patients proceeded with other surgical options. A VNS was also implanted as a further palliative measure in 2 patients with nonlesional epilepsy in whom, despite a reduction in drop attacks, seizures remained generalized or had bilateral onsets after corpus callosotomy (Engel IB, follow-up 100 months; Engel IIIA, follow-up 50 months). In 1 case, a VNS device was used as a supplementary palliative treatment following a temporal lobe resection in the patient with the SCN1A mutation (Engel IIIA, 102-month follow-up). Use of AEDs Three of the 29 patients who underwent resection achieved long-term seizure freedom without the use of any AED (follow-up 91, 79, and 62 months), and 2 more remained seizure free with continuation of monotherapy (follow-up 77 and 42 months). Eight patients discontinued 1 or 2 AEDs after surgery, and 14 of the 29 patients remained on polytherapy. Two patients who underwent resection were lost to follow-up and their current AED dose was not available. In the patients who underwent palliative treatments, there was no benefit with regard to AEDs. Autism Severity and Surgical Outcome Patients with less severe autism, specifically those with autism severity scores of 1 or 2 (10 patients total) all achieved seizure outcomes of Engel I or II, whereas all 21 patients with Engel outcome III or IV were among the 43 patients with severe ASD (severity score of 3). This trend approached but did not reach significance (p = 0.080). Functional and Behavioral Outcomes Aberrant Behaviors Observed in the Clinical Setting In a direct examination by a clinician, improvements in aberrant behaviors were noted in 24 of the 27 patients who, prior to surgery, presented with significant levels of aggression and other disruptive behaviors, including 20 cases of marked behavioral improvement in patients with severe ASD, and 7 cases of such improvement in patients who did not become seizure-free. Seventeen of the patients who experienced marked improvement had resective treatment (1 removal of a hypothalamic hamartoma, 3 temporal lobectomies, 4 frontal resections, and 9 multilobar operations [including 2 hemispherotomies]), and 7 had undergone corpus callosotomy. Two callosotomy patients who initially improved later experienced behavioral regression brought on by recurrence of seizures. Long-Term Functional and Behavioral Outcomes Progress was noted over time in the majority of patients who experienced seizure reduction following surgery (Table 3). According to caregivers reports, a significant number of patients experienced gains in cognitive function (e.g., increased awareness, ability to pay attention and retain new information), which the parents partially attributed to a reduction in AED side effects. Many patients also made progress in the development of language and other communication skills (e.g., eye contact, using gestures to communicate, more appropriate response to caregivers mood and facial expressions); social behavior (e.g., more engagement with caregivers, ability to play cooperatively with a sibling, initiative to hold hands with children at school); and ability to cope with changes in personal routines and/or fixations and the environment. Among other postoperative trends noted were improvements in overall well-being (e.g., sleep quality, digestive health, physical activity throughout the day) and reduction in concerns over patients safety and in the amount of daily care required. Notably, the fewest negative changes were observed in socialization, expressive language, and nonverbal communication, areas particularly problematic in ASD. The severity of all behavioral symptoms observed by caregivers on a daily basis improved in 25 of the 46 patients evaluated, whereas 21 reported no change or worsening of behavior (Table 3). Also, some parents who noted overall improvement reported negative symptoms as well 202 J Neurosurg Pediatr Volume 19 February 2017

8 Autism and epilepsy surgery (e.g., hyperactivity, inability to control provoked aggression, and other impulsive behaviors). There was no significant association with Engel outcome, with approximately half of the parents of patients with Engel I outcomes (9 of 16) reporting behavioral improvement, and half of the parents of patients with Engel outcomes II IV reporting behavioral improvement (16 of 30), with variability in individual outcomes. In some cases, a moderate reduction in seizures was associated with marked improvements in behavior, obvious to all observers (e.g., Case 39). In other cases some behavioral symptoms persisted in patients who became seizure free and made excellent developmental progress (e.g., Case 11), as well as cases in which aberrant behaviors ceased after corpus callosotomy, but returned 1 year later following recurrence of seizures (e.g., Case 49). Additionally, parents of 3 surgical patients with severe ASD reported no noticeable change in behavior or overall functional status resulting from seizure reduction (Case 5: Engel IIB, 45-month follow-up; Case 13: Engel IIIA, 102-month follow-up; Case 22: Engel IIIA, 35-month follow-up). The 2 worst functional-behavioral outcomes in our series, with no reported long-term improvement in any area compared with baseline and worsening of some symptoms were associated with Engel IV outcomes (Case 7, 102-month follow-up, and Case 37, 47-month follow-up). Overall Satisfaction With Outcome Forty-three of the 49 families with recent follow-up (88%) considered the outcome of surgical treatment positive, with improvements in quality of life and cognitive ability most frequently cited as the main benefits of improved seizure control (Table 4), although 1 family in this group also stated that they had hoped for a better outcome with regard to the child s physical development, and 2 families were disappointed that their children remained nonverbal. Parents of 3 children who experienced seizure reduction but did not become seizure free attributed better postsurgical seizure control to continued adjustments in medical therapy. Surgery-Related Complications Four of the 56 patients in our report experienced complications related to surgical treatment. Three of these were associated with subdural electrodes and 1 with a lesionectomy without invasive monitoring. The complications included 2 cases of hydrocephalus (1 required placement of a ventriculoperitoneal shunt, and 1 resolved after placement of an external ventricular drain), and 2 cases of infection necessitating treatment with intravenous antibiotics (purulence, or stain or culture positive for Staphylococcus aureus noted at the time of electrode removal). Other than 3 cases of expected peripheral visual field loss after occipital resections, no permanent surgery-related physical or neurological deficits were observed. There were no complications related to the callosotomy and VNS procedures. All surgical treatments were successfully performed according to clinical plan, including invasive monitoring, and the complications were deemed unrelated to the presence of autistic symptoms or behaviors. This success was credited partially to enhanced inpatient care protocols focused on minimizing stress to the child and expanding direct family involvement, including intensive involvement with childlife specialists and music therapists, anesthesia protocols involving family members and familiar companions, multidisciplinary family meetings, and familiarity with the epilepsy monitoring unit environment and staff (our unpublished data). Discussion Our results support the feasibility of epilepsy surgery in children with ASD, highlight the potential for improved seizure control and associated cognitive-behavioral gains, and attest to the safety of surgical intervention in this population. 21,37,38 Contrary to the prevalent concern that epilepsy surgery in patients with ASD may have limited utility, 10,11,14,40,45 we found that even for severely affected individuals, surgery often had tangible benefits, reducing the burden of seizures and aberrant behaviors on the families and facilities that provide lifelong care to these patients, many of whom require special medical attention and very substantial support due to the presence of developmental delay and other comorbidities. Historical Perspective Although autism was historically thought to be a general contraindication to epilepsy surgery, the initial case reports of surgical intervention related positive outcomes. Gillberg et al. reported on 2 boys with severe autism and epilepsy associated with TS pathology, both of whom became seizure free after resection of epileptogenic foci. 20 One of these patients experienced a drastic behavioral improvement and ultimately loss of autism diagnosis. In another early report of 2 surgical cases involving the temporal lobe, long-term seizure freedom was achieved by both patients, and autistic features demonstrated a clear response to surgical treatment. 33 Later reports, however, each including 2 5 patients with autism and intractable seizures, described mixed surgical results ranging from seizure freedom to seizure worsening, with no consistent effect on the severity of behavioral symptoms. 8,9,17,29,31, 32,45 Furthermore, McLellan et al. found that among 23 patients with autism who underwent temporal lobectomy, 10 became seizure free (with 2 cases of loss of autism diagnosis), but in their series children without psychiatric comorbidities or with psychiatric disorders other than autism had better overall seizure control postoperatively. 30 Larger studies of surgical outcomes in patients with ASD are not found in the literature, with the exception of 2 nonconsecutive series of patients who underwent VNS placement, the majority of whom experienced seizure reduction in the 1st year after surgery, according to data voluntarily submitted to the manufacturer by the treating physicians. 27,35 However, this trend was not confirmed in 8 consecutive independently reported patients with autism in whom implantation of a VNS had no appreciable benefit. 10 With the exception of autistic patients with epileptogenic tubers, 8,40,47 the outcomes of epilepsy surgery in children with intractable seizures and autistic symptoms are not well established in the literature. Presurgical Counseling In evaluating an autistic patient for epilepsy surgery, we J Neurosurg Pediatr Volume 19 February

9 M. A. Kokoszka et al. used an approach similar to that taken with other pediatric patients with medically refractory seizures in the risk-benefit analysis, and multiple adjustments in pharmacotherapy were undertaken prior to considering surgery. Most of the patients in our series were severely autistic, some also had other chronic comorbidities, and the 7 patients in our group who underwent surgery when they were older than 17 years required a pediatric setting in which practitioners had expertise in managing patients with severe developmental delay and experience in performing presurgical evaluation and invasive monitoring in patients with a history of uncooperative behavior. However, unlike the parents of children with MRE and global developmental delay, who are typically focused on preventing the detrimental effect of uncontrolled seizures and high doses of AEDs on the developing brain, in our experience, the parents of children with the comorbidity of autism seek additional answers as to the relative implications of ASD and seizures for the child s development and functional-behavioral status, and the outcomes of epilepsy surgery in autistic patients. The literature cites poor long-term outcomes in children with autism and uncontrolled seizures who did not undergo surgery, 7,9,24 but offers little insight into factors predictive of surgical outcome in this diverse population. Because autism is presumed to arise from a combination of environmental and genetic factors, the unknown implications of these presumed risk factors on seizure outcomes were part of the presurgical discussion with families and caregivers, and genetic counseling was offered. Availability of genetic testing was often limited prior to surgery, either due to historical reasons or the high cost associated with comprehensive autism and epilepsy panels, which insurance providers did not consider justifiable, given the general lack of evidence linking specific genetic findings to surgical prognosis. All of our patients with either Engel III or Engel IV outcome had severe ASD, and although it is important to set realistic expectations and keep in mind that such cases may be challenging and that many of the severely affected patients have limited potential for full resolution of all developmental symptoms, surgery is always considered in patients with pharmacoresistant seizures who are potential surgical candidates at our epilepsy center, regardless of the presence of ASD. Resections and Seizure Outcomes As with the general population of pediatric patients with epilepsy who undergo surgery, the best seizure freedom rates in our series were observed in hemispherotomy cases, 5,41 followed by focal and multifocal resections 15,16,44 (Fig. 1), but in our experience, localizable but nonlesional epilepsy cases were the most challenging. Recent metaanalyses of outcomes of resections in pediatric epilepsy patients found long-term seizure freedom rates of 60% 81% associated with MRI abnormalities, and 49% 51% in patients with normal MRI findings. 15,16 Thirteen of the 20 patients in our series with MRI abnormalities became seizure free after resection (10 of 14, or 71% of patients with > 24-month follow-up), compared with 2 of the 9 patients (1 of 7, or 14% of patients with > 24-month follow-up) with normal MRI findings who achieved an Engel I outcome. Similar to the general population, for the patients in our series who experienced seizure recurrence after surgery, the cause of relapse was often presumed to be genetic, especially in cases in which the resected tissue appeared morphologically normal on pathological assessment, such as in the patient in Case 23, who was later revealed to have a PCDH19 mutation. Similarly, in 2 patients with abnormal MRI and concordant EEG findings (Case 40, focal temporal lobe epilepsy with MTS; and Case 19, vascular etiology), both of them experienced new-onset generalized seizures after > 2 years of seizure freedom. Genetic testing was repeated after surgery, and in the patient in Case 40 a novel mutation in the PRICKLE1 gene was diagnosed. Our results are also consistent with recent reports of poor outcomes of surgery in patients with SCN1A mutations. 4,42 On the other hand, the patient in Case 10, with a family history of epilepsy associated with developmental delay (who could not be positively diagnosed using the genetic testing available at the time) became seizure free and made excellent developmental progress after a resection. Although many of the patients in our series were able to reduce the number of medications following surgery, only 3 of the 22 patients who underwent resection with a > 2-year follow-up were able to discontinue all AEDs (Cases 10, 32, and 33). This was partially due to occurrence of breakthrough seizures, the presence of EEG abnormalities despite seizure freedom, and general parental reluctance to discontinue medications, leading to a conservative pace in weaning patients off AEDs. One of the 3 patients who did not have a resection following invasive monitoring has not had a seizure in the 6 years since implantation of intracranial electrodes, and the other 2 patients have had a drastic improvement in their epilepsy. Similar cases of seizure remission after invasive monitoring have been previously reported in a minority of patients with both lesional and nonlesional etiology, 26,38 and this may be at least in part due to mechanical modulation caused by surgical intervention, leading to the disruption of epileptogenic networks. Outcomes of Palliative Procedures Notably, more than half of patients in our series (32 of 56) had one or more palliative procedures. The relatively high number of callosotomies performed (26 in 56 patients) reflects a high incidence of generalized epilepsy and atonic seizures in our group. At our center, corpus callosotomy remains the recommended treatment for atonic seizures with or without a Lennox-Gastaut pattern. Our results support existing data showing that callosotomy can be performed safely with minimal risk of a new deficit. 36 The procedure brought seizure relief to 21 of 26 patients; either directly, by ameliorating atonic and/or tonic seizures, and/or generalized seizures with apneic-cyanotic spells, or indirectly, by enabling lateralization and resective treatment. 34 Our experience in some patients with nonlesional epilepsy that lateralized after callosotomy, but whose seizures recurred after resective procedures from another location, has prompted a shift toward the possible application of responsive neurostimulation (RNS) in similar future cases. Five patients in our series had a VNS device implanted prior to callosotomy, and 3 of the patients who ultimately underwent resective procedures also initially opted for 204 J Neurosurg Pediatr Volume 19 February 2017

10 Autism and epilepsy surgery VNS implantation before considering other surgical modalities that were better indicated. Such decision making reflects a natural and inherent bias favoring a reversible extracranial procedure over an irreversible resective or disconnective intracranial operation, and in our experience is not unique to the autistic population. Currently, 5 of the 14 patients who underwent VNS placement are experiencing some relief from seizures due to VNS combined with pharmacotherapy, and are not seeking other treatment options. Future Directions: Neuromodulation in Lieu of Ablation Our patients who underwent resection had less favorable outcomes in cases in which apparent focal or regionalized epilepsy led to lobectomies and/or disconnections, compared with complete remission of seizures in the hemispherotomy group. Where MRI findings were nonlesional, only 2 of 9 patients achieved an Engel I outcome, compared with 2 patients with Engel II and 5 with Engel III seizure control. In the 5 cases in our series in which complete callosotomy was associated with the opportunity for a more focal or lateralized attempt at seizure control via staged resection and/or disconnection, our results were not curative, with Engel II and III outcomes. These findings have resulted in a change in approach that we have taken toward patients with autism and nonlesional focal or regional epilepsy, who are not candidates for hemispherotomy. We are concerned about more global hemispheric or even bihemispheric epileptogenic potential in these patients, that with time can result in early and late failures. Thus, we have begun to recommend a modulatory rather than ablative approach in similar patients, with preliminarily beneficial results (our unpublished data). Until a better understanding of the mechanisms of seizure propagation has been established through genetic and epileptic network analyses, the RNS procedure using strip electrodes in the region of putative seizure onsets, combined with anterior thalamic nucleus electrodes, has been proposed as a palliative measure. Given the ability to perform long-term electrocorticographic analyses with an RNS device, we feel that we will have better diagnostic and potentially therapeutic advantages to guide long-term management. Outcomes of Surgery With Regard to Symptoms Associated With ASD Rare cases of the loss of autism diagnosis have been previously described in some autistic patients undergoing surgery for intractable seizures. 20,30 While not observed in our series, improvements in aberrant behaviors such as violence and defiant behavior were among the benefits of surgery, suggesting that at least some of the behavioral issues prior to surgery were related to poorly controlled seizures. Although many of the gains reported here may be intangible when measured by standard methods, the trend toward overall cognitive-behavioral improvement observed in this series is in stark contrast to the cognitive decline and overall poor outcomes reported in children with ASD and refractory seizures. 7,9,18,23,24 According to caregivers reports, the reduction in aberrant behaviors was particularly beneficial to family life, where unprovoked aggression and other disruptive behaviors can be more of a concern than the social delay, restricted interests, or repetitive behaviors of autism. 28 Study Limitations Our study is hampered by a number of limitations. The behavioral changes reported here were assessed using both neuropsychological records and nonstandard methods. We chose this direct caregiver perspective to assess behavioral outcome as a meaningful and sensitive measure of the patients functional status, since the changes were noted by those providing daily care and support to the patients, and we did not believe that a more globally applicable grading scale was available in the population we were studying, without prospective planning. The study is limited by retrospective design, and no control group is available for comparison of outcomes because clinically indicated surgical treatments were provided to all patients. Although uncontrolled seizures in autism correlate with poor longterm outcomes, spontaneous remission of epilepsy has also been reported, 9 and the possibility cannot be definitively excluded. Our numbers do not allow for statistical analysis within each clinical category, but instead build on data from other studies with similar designs 4,30,32,40,42,45,46 to provide a more complete picture of the range of underlying pathologies and outcomes of epilepsy surgery in this genetically and etiologically diverse population who all share the diagnoses of epilepsy and ASD. The association between autism and epilepsy continues to be the subject of intense scientific inquiry, with a rapidly evolving understanding at genetic, synaptic, and molecular levels. Surgical and behavioral failures in our population highlight the limited understanding of this association and the imperfect ability to identify the optimal surgical candidate. However, we know that favorable surgical outcomes are possible in autistic patients, and the autistic population cannot be excluded when medical intractability dictates the need for intervention in epilepsy that is believed to be surgically remediable. Conclusions Patients with autism and pharmacoresistant epilepsy can achieve worthwhile seizure reduction and should not be denied surgery. However, much remains unknown about the etiology of autism, so genetic counseling should be provided regardless of MRI findings, conservative adjustments in pharmacotherapy are recommended, and novel surgical methods for the understanding of epileptogenic networks and their modulation such as RNS should be considered. Continued review of epilepsy surgery outcomes in autism is needed as new genetic and other risk factors emerge and may affect prognosis. Acknowledgments We thank Dr. Robert R. Goodman for his input during early stages of the study. The support for this study was provided entirely by the Department of Neurosurgery, Mount Sinai Health System, New York, NY. References 1. American Psychiatric Association: Diagnostic and Statisti- J Neurosurg Pediatr Volume 19 February