Hemispherectomy is an effective treatment option for children with medically refractory epilepsy caused by extensive congenital or acquired hemispheric lesions. Reported seizure freedom rates after hemispherectomy vary from 52% to 80%, and multiple seizure outcome indicators have been identified, providing valuable information to counsel families about seizure outcome (Kossoff et al., 2003; Jonas et al., 2004; Delalande et al., 2007; Hallbook et al., 2010; Moosa et al., 2013). On the other hand, family counseling for functional outcome remains difficult owing to lack of good data on functional status after hemispherectomy. Families faced with a child requiring hemispherectomy often ask: Can the hemiplegia get worse? Can he walk unaided? How does the visual field defect affect daily life? Will she talk and read like her peers? What are the behavior and mental health challenges? Will he attend a mainstream school and work independently as an adult? Few studies have attempted to address these functional outcomes by reporting either general observations or using a battery of neuropsychological tests (Maehara et al., 2002; Pulsifer et al., 2004; Korkman et al., 2005; Lettori et al., 2008; Liegeois et al., 2008); however, none looked at the long-term functional outcomes using a “patient-centered approach” (Wiebe & Berg, 2013).
Two large series on hemispherectomy focused primarily on seizure outcome also reported on the motor (Kossoff et al., 2003; Jonas et al., 2004) and language outcomes (Jonas et al., 2004). However, in one study, comprehensive scoring of the motor strength of the hemiplegic limbs was examined; the effect of this on the patient's ability to walk was not reported (Jonas et al., 2004). Various other studies addressed the impairment or neurologic deficits following hemispherectomy rather than true limitation of function in daily activities (Maehara et al., 2002; Pulsifer et al., 2004; Korkman et al., 2005; Lettori et al., 2008; Liegeois et al., 2008). Furthermore, functional outcome predictors remain largely unexplored in these previous studies (Maehara et al., 2002; Devlin et al., 2003; Kossoff et al., 2003; Jonas et al., 2004; Pulsifer et al., 2004; van Empelen et al., 2004; Korkman et al., 2005; Lettori et al., 2008; Liegeois et al., 2008).
The goals of this study are the following: (1) to examine the functional outcome after hemispherectomy in the key functional domains of daily life—ability to walk, visual symptoms, spoken language skills, reading ability, behavior, and academics/employment—“as perceived by the families and patients”; and (2) to identify the prognostic factors that determine the functional outcomes.
Materials and Methods
We reviewed the medical records of all patients who underwent hemispherectomy between January 1997 and June 2009 at our center; patients age 18 or younger at the time of surgery were included in the study. Preoperative clinical, electroencephalography (EEG), video- EEG, radiologic, and surgical data were collected. Preoperative neurologic deficits including presence and severity of motor deficits, ability to walk, language skills, and visual deficits were documented. When a child was younger than the expected age for development of a particular function, or when testing was not objectively reliable and possible, their preoperative status was classified as “indeterminate.” For example, the hemiparesis status was categorized as “indeterminate” in young infants who had neurologic signs to suggest lateralized pyramidal dysfunction but objective quantification of weakness was not possible. Similarly, language status was categorized as “indeterminate” in infants and young toddlers where testing was unreliable and wide variability of normal language skills exist.
Postoperative functional outcome
Functional outcome was assessed using a structured questionnaire inquiring about functional (ambulation, visual symptoms, spoken language and reading ability, behavioral, and academic/employment) and seizure outcomes (Data S1). Each question had simply formulated multiple responses from which the families chose the one that best described the status of the patient from a real life perspective. Because the age of the patients in the study ranged from 2 to 28 years at the time of posthemispherectomy functional evaluation, we allowed options in the answers to include age appropriateness. For example, for the verbal language skills, a child age 2 years with a vocabulary of >50 words and ability to combine two words together would be categorized as “appropriate for age.”
The questionnaire was initially mailed and the families were contacted over the phone. After obtaining informed consent, the questionnaire was completed over the phone. Parents or guardians, and in some instances patients themselves, completed the survey. Few families mailed us the completed questionnaire. When we were unable to contact the families by phone or mail, we attempted to reach them through e-mail (when available) to complete a secure online survey using the REDCap electronic data capture tool hosted at our center.
For the analysis of prognostic predictors, functional outcome for ambulation was classified into three groups: (1) independent walking, (2) able to walk few steps with assistance, and (3) unable to walk. The spoken language, reading, and behavioral outcome statuses were categorized as good or poor. Spoken language outcome was dichotomized to good (age appropriate or just below age level but not limiting communication in daily life) or poor outcome (severe delay or 2–3 word phrases only or unclear words or nonverbal). Similarly, reading outcome was dichotomized to good (age appropriate reading or few years below age/grade level but able to read sentences) or poor outcome (able to read only familiar words or alphabets or unable to read). For behavior, good outcome referred to children with no or minimal behavioral problems not affecting the life of the child and family.
Descriptive statistics were used for each variable including means and standard deviation for continuous variables and frequencies for categorical variables. First, the data were analyzed using Wilcoxon rank-sum, chi-square, and Fisher's exact tests to compare between individual favorable versus unfavorable functional outcomes. This initial analysis provided data for potential prognostic factors. Variables with a significance level of 5% on univariate analysis were then tested in a multivariate linear regression model where statistical significance was considered at the 5% level. Multi-colinearity was addressed and controlled for when appropriate. These methods allowed testing the correlation of specific variables with outcome while taking into account any interactions and associations among variables.
The institutional review board at the Cleveland Clinic approved the study.
Of 186 patients who had hemispherectomy during the study period, 125 completed the questionnaire; one patient refused to participate in the study and the remaining 60 patients could not be reached primarily due to change in their contact information, many being out of state or international patients. Of 125, we excluded 10 patients as they had new nonepileptic spells on follow-up without definite evidence of epileptic seizures on video-EEG monitoring. Mean follow-up was 6.05 years (standard deviation [SD] 3.1, median 5.25 years; range 1.33–13.5 years). Mean age at follow-up was 12.73 years (SD 6.02, median 11.42 years; range 2.25–28.6 years). There were no differences between survey responders and nonresponders regarding age at surgery, age at epilepsy onset, gender, magnetic resonance imaging (MRI) findings, or baseline language performance. There was, however, a tendency for better survey response rates in patients whose seizures had recurred after surgery: 62% responded in the seizure-free group, compared to 78% in seizure-recurrence group (p = 0.04).
Preoperative neurologic deficits and baseline characteristics of the cohort are shown in Tables 1 and S1. Ninety-eight children (85%) were age 3 years or younger at the time of epilepsy onset, with 64 being younger than 1 year of age. Daily seizures despite multiple antiepileptic medications were noted in 88 patients (76.5%); 21 (18.2%) had at least one seizure every week, 4 patients (3.5%) had an average of 1–3 seizures per month; 2 patients (1.7%) had recurrent refractory status epilepticus every few months.
Table 1. Preoperative neurologic deficits in the cohort (n = 115)
|Motor deficits|| ||Ambulation status|| |
|Severe hemiplegia||74 (64)||Walk unaided||71 (62)|
|Mild to moderate hemiplegia||24 (21)||Nonambulant, <18 months old||31 (27)|
|Indeterminatea||10 (9)||Unable to walk, >18 months old||9 (8)|
|Bilateral motor deficitsb||6 (5)||Few steps with assistance||4 (3)|
|No hemiparesisc||1 (1)|| || |
|Spoken language skills|| ||Visual deficits||49 (42)|
|Delayed||70 (61)||Contralateral homonymous hemianopic defect||46 (40)|
|Normal||15 (13)||Unable to evaluate reliably||3 (3)|
|Indeterminated||30 (26)||Global visual dysfunction||17 (15)|
| || ||No visual field defects|| |
Postoperative seizure outcome
At last follow-up, 70 patients (61%) were seizure-free since surgery. Of the 45 with seizure recurrence, 8 (7% of 115) achieved late remission (seizure-free for 1 year or more at last follow-up) and 15 (13% of 115) had >90% decrease in seizure count accounting for favorable outcome in 93 patients (81%) from the entire cohort (n = 115). Ten patients (8.6% of 115) reported 50–90% reduction in seizure frequency and 12 (10.4% of 115) patients had no or minimal improvement in seizure frequency accounting for unfavorable outcome in 22 (19%) of patients out of the entire cohort. Of 70 seizure-free patients, 58 were no longer on antiseizure medications. Longitudinal seizure outcome in the original cohort of 186 patients has been reported previously (Moosa et al., 2013).
Postoperative functional outcome
Of the 115 patients, 24 were aged 18 years or older; 81 were between 6 and 17 years, and 10 patients were between 2 and 5 years at the time of functional outcome evaluation. The functional outcome of 115 patients is shown in Table 2. Information about reading and school/academic abilities were not inquired about in 10 children who were younger than 6 years. Of 81 patients aged between 6 and 17 years, the majority required assistance at school. Among the 24 adults, only 5 were gainfully employed and independent. All five employed adult patients had right hemispherectomy; three patients had Rasmussen encephalitis, one had stroke, and the other had hemispheric dysplasia as etiology; four of these five patients were seizure-free since surgery.
Table 2. Functional outcome after hemispherectomy in the cohort (n = 115)
|Walk independently||96 (83.4)||Unchanged||62 (54|
|Few steps with assistance||10 (8.6)||Worsea||42 (36.5)|
|Unable to walk—older than 18 monthsb||9 (7.8)||Betterc||11 (9.5)|
|Age appropriate||39 (33.9)||No behavioral issues||62 (53.9)|
|Mild impairment||41 (35.7)||Minimal problems at home with no significant impairment||22 (19.1)|
|2–3 word phrases only||18 (15.7)||Significant problems affecting school/social life||19 (16.5)|
|Few unclear words||8 (6.9)||Require constant supervision due to behavioral problems||12 (10.4)|
|Nonverbal||9 (7.8)|| || |
|Newly recognized visual defects, not present pre-op||28 (24.3)||Age appropriate||19 (18.1)|
|Unsure if new or preexisting||15 (13.1)||Few years below age level||25 (23.8)|
|Unchanged from pre-op||19 (16.5)||Reads familiar words or picture naming only||27 (25.7)|
|Minimal—not affecting daily activities||30 (26.1)||Knows alphabets and or numbers only||15 (14.3)|
|“No visual symptoms reported”||23 (20)||Cannot read||19 (18.1)|
|Mainstream without assistance||5 (6.2)||Employed gainfully||5 (20.8)|
|Mainstream with assistance||48 (59.2)||Attends day workshop||9 (37.5)|
|Special school for disabled||22 (27.2)||In supervised facility/school with assistance||10 (41.7)|
|Home cared/minimally functional||6 (7.4)|| || |
Predictors of functional outcome
On univariate analysis, structural abnormalities in the nonoperated hemisphere on MRI, preexisting bilateral motor deficits, and postoperative seizure recurrence were associated with inability to walk independently (Table 3). On multivariate analysis, all three factors were independently predictive of inability to walk. Factors that did not affect ambulatory outcome status include: age at surgery, age at epilepsy onset, etiology of epilepsy, epilepsy duration (“seizure onset-to-surgery” interval), bilateral abnormalities on the fluorodeoxyglucose-positron emission tomography (FDG-PET) and preoperative ambulatory status. Patients with Rasmussen encephalitis were more likely to report “worsening” of the hemiparesis after surgery; 7 (70%) of 10 patients with Rasmussen encephalitis experienced “worsening of hemiplegia” as opposed to 35 (33%) of 105 patients with all other etiologies (p = 0.02). However, etiology had no effect on the ambulation ability after hemispherectomy. All 10 young infants with “indeterminate” preoperative hemiparesis achieved independent walking on follow-up.
Table 3. Univariate analysis of predictors of long-term motor outcome after hemispherectomy
|Seizure freedom status|| || || || |
|Seizure-free (70)||66 (94)||2 (3)||2 (3)||0.0005|
|Seizure recurrence (45)||30 (66)||8 (18)||7 (16)|
|MRI abnormalities on the nonoperated hemisphere|| || || || |
|Normal (72)||67 (93)||3 (4)||2 (3)||0.01|
|White matter abnormalities only (16)||12 (75)||3 (19)||1 (6)|
|Cortical change in ≤1 lobe (13)||10 (77)||1 (7.6)||2 (15.4)|
|Cortical changes in ≥2 lobes (11)||5 (45.4)||2 (18)||4 (3.6)|
|Unclear extent (3)||2 (67)||1 (33)||0|
|Preoperative motor deficits|| || || || |
|Severe hemiplegia (74)||64 (86.5)||5 (6.75)||5 (6.75)||0.001|
|Mild to moderate hemiplegia (24)||19 (79)||5 (21)||0|
|Bilateral motor deficits (6)||2 (33)||0||4 (67)|
|Indeterminate (10)||10 (100)||0||0|
|No hemiparesis (1)||1 (100)||0||0|
Spoken language skills
Variables affecting the postoperative spoken language ability on univariate analysis are shown in Table 4. Factors that did not affect verbal language outcome include: age at surgery, duration of epilepsy prior to surgery, etiology of epilepsy, and bilateral abnormalities on the FDG-PET. A trend toward poor verbal language outcome was noted in children with epilepsy onset at younger age, and in children who had left hemispherectomy, but the differences were not significant (Table 4). We further studied the age of epilepsy onset and age at surgery in patients with left hemispherectomy and found no particular pattern predictive of spoken language outcome in this cohort.
Table 4. Univariate analysis of predictors of long-term language and reading ability after hemispherectomy
|Mean age at seizure onset (years)||1.94 (±2.61)||1.07 (±2.26)||0.06||2.53 (±3.1)||1.26 (±1.97)||0.01|
|Side of surgery (%)|| || || || || || |
|Left||32 (61.5)||20 (38.5)||0.09||19 (39.6)||29 (60.4)|| NS|
|Right||48 (76)||15 (24)||25 (43.9)||32 (56.1)|
|Preoperative|| || || || || || |
|Language skills (%)|| || || || || || |
|Delay||48 (68.6)||22 (31.4)||<0.04||25 (36.2)||44 (63.8)||<0.03|
|Normal||14 (93.3)||1 (6.7)f||11 (73.3)||4 (26.6)|
|Indeterminate||18 (60)||12 (40)||8 (38.1)||13 (61.9)|
|MRI abnormalities in the nonoperated hemisphere (%)|| || || || || || |
|Normal||58 (80.6)||14 (19.4)||0.001||33 (50)||33 (50)||0.04|
|White matter changes only||8 (50)||8 (50)||3 (21.4)||11 (78.6)|
|Cortical changes ≤1 lobe||8 (61.5)||5 (38.5)||6 (50)||6 (50)|
|Cortical changes in ≥2 lobes||3 (27.3)||8 (72.7)||1 (10)||9 (90)|
|Unclear extent||3 (100)||0||1 (33)||2 (67)|
|Seizure freedom status (%)|| || || || || || |
|Seizure-free||57 (81.4)||13 (18.6)||0.0006||35 (57.4)||26 (42.6)||0.0001|
|Seizure recurrence||23 (51)||22 (49)||9 (20.5)||35 (79.5)|
On multivariate analysis, structural abnormalities in the nonoperated hemisphere on MRI, children with preoperative language delay or indeterminate language status (young infants), and postoperative seizure recurrence (odds ratio [OR] 3.18, 95% confidence interval [CI] 0.99–10.19, p = 0.04) emerged as independent risk factors for poor verbal language ability at last follow-up (Table 4). Children with multilobar abnormalities on the contralateral hemisphere MRI had the worst spoken language outcome (OR 13.98, 95% CI 2.77–90.37, p = 0.001) as opposed to children with normal contralateral hemisphere. Better language skills at last follow-up were seen in children with normal preoperative language skills when compared to those with preoperative indeterminate language status (OR 11.16, 95% CI 1.63–230.57, p = 0.01) or language delay (OR 4.6, 95% CI 0.73–91.93, p = 0.1).
We inquired on the reading ability of 105 children aged 6 years or older at the time of functional outcome evaluation (Table 2). Of 105 patients, 58% had poor reading skills. On univariate analysis, reading ability was adversely affected by younger age at epilepsy onset, structural abnormalities in the nonoperated hemisphere on MRI, verbal language delay prior to surgery, and seizure recurrence (Table 4). Factors that did not affect reading ability include: age at surgery, duration of epilepsy prior to surgery, etiology of epilepsy, and bilateral abnormalities on the FDG-PET. After multivariate analysis, younger age at epilepsy onset, cortical abnormalities in the contralateral hemisphere on MRI (OR 24.07, 95% CI 2.24–764.73, p = 0.005 for patients with multilobar abnormalities as opposed to normal opposite hemisphere), and seizure recurrence after hemispherectomy (OR 5.00, 95% CI 1.55–18.59, p = 0.006)remained independent predictors of poor reading outcome.
Seizure recurrence after hemispherectomy was the only factor predictive of poor behavioral outcome. Significant behavioral problems were reported in 18 (40%) of 45 patients with seizure recurrence as opposed to 12 (17%) of 70 with seizure freedom (p = 0.02). A trend toward poor outcome in children with epilepsy onset at younger age was also noted. Mean age at epilepsy onset in patients with significant behavior problems was 1.04 years (±1.68) as opposed 1.9 years (±2.74) in those with minimal or no behavioral problems (p = 0.07). Age at surgery, duration of epilepsy prior to surgery, etiology of epilepsy, and bilateral abnormalities on MRI or FDG-PET had no impact on the behavioral outcome.
In this cohort of 115 children, at a mean follow-up of 6.05 years after hemispherectomy, 83% patients walked independently, 73% had minimal or no behavioral problems, 69.5% had satisfactory spoken language skills, and 42% had good reading skills. In 76% of patients, limitations related to visual field deficits were not perceived as significant. Two factors emerged as major predictors of poor outcome in ambulation, spoken language, and reading abilities: (1) seizure recurrence and (2) contralateral hemisphere abnormalities on MRI. In addition, children younger than <18 months with indeterminate language status at surgery also had poor postoperative spoken language skills, and younger age of epilepsy onset predicted poor reading skills. Other variables such as etiology of epilepsy and “epilepsy duration” had no effect on the functional status in this cohort.
Seizure freedom after surgery emerged as the key factor that favorably impacted outcome in all functional domains studied. This is similar to findings observed in other studies of children (Maehara et al., 2002; Ramantani et al., 2013). We excluded patients with nonepileptic spells on follow-up because we intended to have two well-defined groups: seizure-free or not, to reliably study the impact of seizure recurrence. Recurrent seizures after surgery adversely affect functional outcome in three possible ways: (1) Deleterious effect of seizures on cognitive function has been shown comparing children with unilateral hemispheric lesions with and without seizures (Vargha-Khadem et al., 1992; Chilosi et al., 2005; Ballantyne et al., 2007). These studies indicate that seizures may “limit the potential of optimum plasticity” of the already injured brain. Therefore, seizure freedom after surgery may relieve the burden of epilepsy on the opposite hemisphere, thereby enabling continued progression of development. (2) In patients with seizure recurrence after hemispherectomy, seizures arise from the nonoperated hemisphere in two thirds of cases (Moosa et al., 2013). Therefore, seizure recurrence after hemispherectomy itself may be a marker of abnormal opposite hemisphere portending to poor outcome. (3) In addition, the adverse effects of antiseizure medications on cognition cannot be disregarded (Hermann et al., 2010). In our series, 58 of 70 patients with seizure freedom were free of antiseizure medications, whereas most patients with seizure recurrence were on two or more medications.
Motor outcome in our patients was similar to another large series of hemispherectomy in children (Kossoff et al., 2003). Serial assessment of strength following hemispherectomy has shown good recovery of strength and tone of proximal leg and arm in almost all patients (van Empelen et al., 2004; Dijkerman et al., 2008). This recovery of proximal motor strength is attributed to compensation from the opposite hemisphere owing to bilateral cortical input for proximal muscles. Hence, in patients with bilateral cortical abnormalities and or bilateral motor deficits, this compensation from the opposite hemisphere may be insufficient and portends to poor motor outcome. In our series, 98% of patients with normal opposite hemisphere on MRI were able to walk independently. Of note, five of the nine patients who were unable to walk in our series were aged between 2 and 5 years at assessment and may still have potential to achieve independent walking.
In our study, 36.5% of patients reported worsening of hemiparesis after surgery, as opposed to 18% (6/33) reported in another study (Devlin et al., 2003). Differences in the severity of preoperative hemiparesis may have accounted for such differences between the two studies: 35 patients in our series had either mild to moderate or difficult to quantify (young infants) hemiparesis preoperatively; the other study did not report the severity of preoperative hemiparesis (Devlin et al., 2003). A small subgroup in our series reported subjective improvement in hemiparesis on follow-up, as noted in other studies (Devlin et al., 2003; Ramantani et al., 2013). Reduction in spasticity along with differential contribution of corticospinal tract projections from the operated and nonoperated hemispheres to the affected limbs may have been responsible for the real or perceived improvement in strength (Govindan et al., 2010; Zsoter et al., 2012; van der Kolk et al., 2013). Two patients in our series lost ambulatory skills after surgery—one patient developed Rasmussen encephalitis in the opposite hemisphere and the other patient had spastic quadriparesis and was barely walking prior to surgery. These events, although rare, need to be considered in preoperative counseling in such clinical scenarios.
Children with multilobar cortical abnormalities in the contralateral hemisphere had the worst spoken language outcome in our study. Similar findings have been reported in another recent study (Boshuisen et al., 2010). Functional outcome after the removal of one hemisphere depends on the balance between the residual critical function at risk for loss in the operated hemisphere, and the integrity and potential plasticity of the opposite hemisphere. The latter is in fact likely to be the major determinant of the long-term functional outcome after hemispherectomy. Therefore, patients with injury to both hemispheres could be left with poor plasticity reserve in the nonoperated hemisphere.
Poor postoperative spoken language outcome in children aged <18 months with indeterminate language skills preoperatively was an interesting finding; these young infants were more likely to have poor spoken language even when compared to the older children with preexisting language delay. This finding opens up the debate between “plasticity” versus “early vulnerability” due to early brain injury. Plasticity proponents argue that brain in the growing phase is considered to be more “plastic,” and hence early injury may be better compensated especially if surgical treatment is offered early (Boatman et al., 1999; Ballantyne et al., 2007, 2008). For example, language deficits due to large infarcts of the left hemisphere are very well compensated if ischemic stroke occurs in early childhood compared to an older child or adult with similar injury (Vargha-Khadem et al., 1992; Ballantyne et al., 2008). On the other hand, proponents of the early vulnerability model have reported poorer cognitive and academic skills in children with early brain injury compared to a similar degree of brain injury in older children (Anderson et al., 2009, 2010). Consequently, poor language skills in these patients may in fact be an index of poor short term memory and verbal intelligence measures rather than a localized dysfunction of the language cortex per se (Liegeois et al., 2008). Several studies have reported correlation between younger age at epilepsy onset and poor intelligence (Vasconcellos et al., 2001; Berg et al., 2012). In our study, there was indeed a trend toward a younger age at epilepsy onset in children with poor spoken language skills. Recent studies on developmental trajectories before and after hemispherectomy have reported that children with poor preoperative developmental quotient were likely to remain in the same category postoperatively (Althausen et al., 2013; Ramantani et al., 2013). Because we did not have the preoperative developmental quotients in all of our patients, we were unable to study this in a similar manner.
Left hemispherectomy in older children and adults is frequently a cause of concern because of potential language areas in the left hemisphere. However, we did not find a significant difference in language outcome between right and left hemispherectomy as in another study (Curtiss et al., 2001). Earlier reports have also indicated that development of language and cognition in infancy requires both hemispheres before the dominance of one (usually left) hemisphere is established. This was supported by reports of poor language outcome in children with prenatal/perinatal injury to the right hemisphere when compared to children with similar injury on the left hemisphere (Liegeois et al., 2008). It has been suggested that the right hemisphere plays a less significant role after the language function is established. In most normal individuals, the left hemisphere is dominant for language function; however, language dominance in patients with large hemispheric lesions with or without epilepsy may not follow this pattern (de Bode & Curtiss, 2000; Ballantyne et al., 2007, 2008). Location and the extent of the brain injury and the age at the time of injury influence the language dominance of the hemisphere. Early severe brain injury to the Broca's and Wernicke's areas in the left hemisphere may result in transfer of language to homologous areas of the opposite hemisphere (Boatman et al., 1999; Chilosi et al., 2005). The oldest age after which such satisfactory language transfer occurred in our series was 8 years. In patients with normal language function despite extensive left hemispheric injury typically indicates preserved function in the opposite hemisphere. However, language lateralization studies should be considered if the extent of injury is localized or the clinical scenario necessitates quantifying the risk of postoperative language deficits.
Of the 105 patients older than 6 years, 58% had poor reading outcome, and this correlated with younger age at epilepsy onset, and seizure recurrence further emphasizing the effect of seizures on the cognition and academic skills. Of 81 school-aged children, the majority required either an individual education plan or enrollment in school for special needs. One earlier report similarly indicated that up to 51% of children with idiopathic or cryptogenic epilepsy may require special educational assistance (Oostrom et al., 2005). Among the 24 adults in our series, 5 were gainfully employed and another 9 were independent but involved in closely supervised jobs. Significant behavioral problems were reported in 27% of patients, and more than one third of these patients warranted constant supervision. The only factor associated with poor behavioral outcome was the seizure recurrence. We did not observe the “burden on normality” affecting the behavior negatively as noted in adults with focal resection (Wilson et al., 2001). Other factors such as antiepileptic medications and parenting style may have affected the behavior problems, but these were not analyzed in our study (Austin et al., 2011).
Two preoperative variables that did not affect the functional outcome—etiology and epilepsy duration—merit further discussion. In an earlier study (Curtiss et al., 2001), etiology of epilepsy affected functional outcome. In this study of 43 patients, the etiology was categorized as developmental (n = 28, including 19 patients with malformations and few with prenatal stroke) and acquired (n = 15, including 10 patients with Rasmussen encephalitis). In our study of 115 patients, there were five etiology categories and we did not differentiate between prenatal and postnatal stroke. There was definitely a trend to better language in Rasmussen encephalitis patients (9 of 10 patients with good language outcome), but this was not statistically significant. Differences in methodology of language assessment also may have accounted for the difference. In general, longer duration of epilepsy prior to surgery is considered as a poor outcome predictor. In our study we did not see such a correlation. It is possible that most hemispherectomy candidates have catastrophic or severe epilepsy necessitating surgery at a very young age and thus have shorter mean epilepsy duration in our cohort. Epilepsy duration may not be an ideal marker for burden of epilepsy in this cohort with early onset frequent daily seizures.
This study has limitations as well as strengths. First, the preoperative assessment of developmental quotient and behavior could not be rigorous due to limitations such as young age, difficulties in testing, frequent seizures, and parenting style (bias). Second, use of a rather “simplistic” questionnaire for ambulatory, language, reading, and behavior assessment may be viewed as a limitation of this study. However, our goal was to gather pragmatic functional outcome from a patient's and family's “real life perspective” so that we could answer the common questions posed by families before and after epilepsy surgery. Third, we were unable to reach 61 patients in the study primarily owing to changes in their contact information. Most baseline characteristics in our study patients were not different from those of the patients who could not be reached except that the patients with seizure recurrence had better survey response rates. In addition, occurrence of autistic features and other behavioral phenotypes (not collected in our study) in the patients may have affected the language and reading skills. Similarly, we were unable to capture the type and intensity of preoperative and postoperative rehabilitation services that could potentially impact long-term outcome. Despite these limitations, we believe the data from this study adds substantial clinically relevant new knowledge on functional outcome from the patients' and their families' perspective. A large number of patients with a long follow-up allowed us to study the previously unstudied determinants of the functional outcome as well. We comprehensively analyzed the various preoperative variables to identify the prognostic factors on functional outcome. In addition, in this outcome study, we focused on the practical limitations in activities of daily life rather than merely quantifying impairment.
In summary, this study shows that impairments affecting the life of patient/family after hemispherectomy, in the order of most to least significance, are reading skills, spoken language, behavior, vision issues, and ambulation. Seizure freedom after hemispherectomy is a strong independent predictor of better functional outcome in all functional domains. Brain MRI abnormalities in the nonoperated hemisphere predict poor motor and language outcomes. Our study will help in counseling of families prior to hemispherectomy using simple and easily comprehensible functional outcome landmarks. In addition, early interventions after hemispherectomy to control seizure recurrence and provide intensive language rehabilitation, when appropriate, may seem prudent to improve long-term functional outcome.