Address correspondence to Tove Hallbook, MD, PhD, Department of Pediatric Neurology, Lund University Hospital, SE – 221 85 Lund, Sweden. E-mail: Tove.Hallbook@telia.com
Purpose: To assess the impact of contralateral magnetic resonance imaging (MRI) findings on seizure outcome after hemispherectomy for refractory epilepsy.
Methods: We retrospectively reviewed 110 children, 0.4–18 (median 5.9) years of age, who underwent hemispherectomy for severe refractory epilepsy at Cleveland Clinic Children’s Hospital. In children with contralateral (as well as ipsilateral) MRI findings appreciated preoperatively, the decision to proceed to surgery was based on other features concordant with the side with the most severe MRI abnormality, including ipsilateral epileptiform discharges, lateralizing seizure semiology, and side of hemiparesis.
Results: We retrospectively observed contralateral MRI abnormalities (predominantly small hemisphere, white matter loss or abnormal signal, or sulcation abnormalities) in 81 patients (74%), including 31 of 43 (72%) with malformations of cortical development (MCD), 31 of 42 (73%) with perinatal injury from infarction or hypoxia, and 15 of 25 (60%) with Rasmussen’s encephalitis, Sturge-Weber syndrome, or posttraumatic encephalomalacia. Among 84 children (76%) with lesions that were congenital or acquired pre- or perinatally, 67 (83%) had contralateral MRI abnormalities (p = 0.02). Contralateral findings were subjectively judged to be mild or moderate in 70 (86%). At follow-up 12–84 (median 24) months after surgery, 79% of patients with contralateral MRI abnormalities were seizure-free compared to 83% of patients without contralateral MRI findings, with no differences based on etiology group or type or severity of contralateral MRI abnormality.
Discussion: MRI abnormalities, usually mild to moderate in severity, were seen in the contralateral hemisphere in the majority of children who underwent hemispherectomy for refractory epilepsy due to various etiologies, especially those that were congenital or early acquired. The contralateral MRI findings, always much less prominent than those in the ipsilateral hemisphere, did not correlate with seizure outcome and may not contraindicate hemispherectomy in otherwise favorable candidates.
The goal of this study was to analyze the implications of contralateral MRI findings on postoperative seizure outcome, to refine the selection criteria for hemispherectomy.
Patients and Methods
From our pediatric epilepsy surgery database, we identified 127 children who underwent hemispherectomy for intractable epilepsy at the Cleveland Clinic Children’s Hospital between 1996 and 2006. We excluded 17 of the 127 children because of recent operation and short follow-up (<12 months) (nine patients), loss to follow-up (six patients), or insufficient archived MRI files for review (two patients). The remaining 110 patients (39 girls) were 0.2–18 (median 5.9) years of age at the time of surgery; 17 (15%) were younger than 1 year of age.
For the purposes of this study, MRIs were retrospectively reviewed blindly by two epileptologists (TH and EW) and a neuroradiologist (PR) for findings ipsilateral and contralateral to the side of surgery. In each case the MRI was performed within a few weeks, usually a few days, before surgery. In the children who had several preoperative evaluations at our center, we used the latest preoperative evaluation for this study.
In every case, the side of surgery was apparent from review of the preoperative MRI, because of the severe, extensive nature of the epileptogenic lesion on one side. We identified and classified all diffuse or focal contralateral MRI changes regarding size of hemisphere, white matter volume and signal, gray matter thickness and signal, gray–white matter differentiation, sulcation pattern, myelination, ventricular size, size of subarachnoidal cerebrospinal fluid (CSF) spaces, and deep gray matter. We subjectively judged the contralateral findings as mild, moderate, or severe. Equivocal contralateral findings were excluded from analysis, including mild sulcation anomalies (five patients) or ventriculomegaly in the presence of a ventriculoperitoneal shunt (seven patients).
MRI images were acquired on a 1.5 T whole body MR (GE, Milwaukee, WI, U.S.A.) scanner. Sagittal and axial T1-weighted images (TR 600 ms, TE 20 ms), T2-weighted (TR 2,800 ms, TE 70 ms), and proton density–weighted (TR 2,800 ms, TE 35 ms) axial and coronal images, both 5 mm thick, were obtained with a 256 × 256 acquisition matrix. In children younger than 2 years of age, T2-weighted and proton density–weighted images were performed, respectively, with TR 2,800 ms, TE 140 ms, TR, 2,800 ms, and TE 70 ms to allow for an evaluation of myelination.
Based on the blinded MRI review, we identified contralateral MRI abnormalities in 81 patients (74%). Children with and without contralateral MRI abnormalities did not differ in age at seizure onset, age at surgery, seizure frequency, number of failed antiepileptic medications, presence of lateralizing signs during seizures, ipsilateral predominance of epileptiform activity on EEG, or duration of follow-up (Table 1). Timing of lesion occurrence appeared to play a role; among 84 children (76%) with lesions that were congenital or acquired pre- or perinatally, 67 (83%) had contralateral MRI abnormalities (p = 0.02). All of the children in both groups had severe medically refractory epilepsy and hemiparesis. Although formal neuropsychological testing was not performed in every case, all of the children were assessed by their pediatric epileptologist, referring child neurologist, and parents to have at least some degree of developmental delay or mental retardation.
Table 1. Patients—clinical features
With contralateral abnormalities (n = 81)
W/o contralateral abnormalities (n = 29)
aSturge-Weber syndrome or encephalomalacia due to trauma, infection, or intrathecal chemotherapy.
bTwenty-six children sustained unprotected falls during seizures; others were nonambulatory but seizures affected postural tone. Fourteen had bouts of status epilepticus (12) or epilepsia partialis continua (2).
cFocal or lateralizing features during seizures (aura, version, focal motor component) were all side-appropriate for surgery. Seizures with no focal or lateralizing features were tonic–clonic, axial tonic, atonic, atypical absence, and spasms.
dWith >70% of epileptiform discharges ictal, interictal, or both ipsilateral to the side of planned hemispherectomy. In the remaining patients with poorly localized EEG, the surgical strategy was based on clinical features, seizure semiology, severity of epilepsy, side of hemiparesis, and MRI findings (16).
eReference: (Peacock et al., 1996; Vining et al., 1997; Alexopoulos et al., 2005; Chapman et al., 2005). Ten patients had a second surgery for persistent seizures after functional hemispherectomy: four with contralateral abnormalities (40%) and two without (20%) had an anatomic hemispherectomy, and three (30%) with contralateral abnormality and one (10%) without had further transection. For the purpose of this study, seizure outcome was assessed following the last procedure.
hMann-Whitney U test [Exact Sig (two-tailed)] was used for nominal and dichotomous variables.
iFisher’s exact test [Exact Sig (two-sided)] was used for interval variables.
Age at epilepsy onset (years)
Age at surgery (years)
Timing of lesion
Pre- and perinatal (<2 m)
Infancy (<2 year)
Childhood (2–18 year)
Lateralizing signs during seizurec
Predominantly ipsilateral EEGd
Hemispherectomy typee (ref)
Hydrocephalus with VPSf
Duration of latest follow-up (mo)
Based on preoperative reports and the available follow-up information, the excluded 17 cases did not appear to differ from the remaining 110 study patients.
Every child in both groups had extensive preoperative evaluation by a pediatric epileptologist at the Cleveland Clinic Children’s Hospital, with video-EEG, MRI, and other testing. Each case was discussed individually at a multidisciplinary epilepsy management conference with input from adult and pediatric epileptologists, epilepsy neurosurgeons, neuroradiologists, neuropsychologists, and bioethicists. In cases with contralateral MRI findings, the clinical decision to proceed to surgery was based on other features concordant with the side with the most severe MRI abnormality, including predominantly ipsilateral ictal and interictal epileptiform discharges (85% of patients), lateralizing seizure semiology (32%), and side of hemiparesis (100%). Although this study does not include patients who were rejected for hemispherectomy, such children in our clinical practice typically lack the localized EEG features or lateralizing seizure semiology, which might counterbalance a contralateral abnormality on MRI.
An attempt was made in every case to obtain MRI, EEG, neuropsychological assessment, and clinical follow-up at 6 months after surgery. Whenever possible, further clinical follow-up and testing as indicated was obtained in clinic or by telephone at yearly intervals thereafter until stable clinical resolution. Follow-up duration did not vary by etiology.
We analyzed the relation between contralateral MRI findings and latest postoperative seizure outcome using Fisher’s exact test [Exact Sig (two-sided)], in the total group, in the different etiology groups, within the contralateral abnormality subgroups, focal/diffuse changes, perceived severity, ipsilateral predominance in ictal or interictal EEG, and lateralizing signs in seizure symptoms. In Table 1 we used Fisher’s exact test [Exact Sig (two-sided)] for interval variables and Mann-Whitney U test [Exact Sig (two-tailed)] for nominal and dichotomous variables. All tests were performed in SPSS 15.0 for Windows (SPSS Inc., Chicago, IL, U.S.A.). p-Values of less than 0.05 were considered statistically significant. The Cleveland Clinic Investigational Review Board approved this study.
Contralateral findings on preoperative MRI
Types of contralateral MRI abnormalities are listed in Table 2. The typical contralateral findings in the MCD group were white matter signal abnormalities, small hemisphere, sulcation abnormalities, white matter loss, and indistinct gray–white junction, as illustrated in Fig. 1A. Typical contralateral findings in the group with infarction/IVH, included white matter signal abnormalities, white matter loss, and ventriculomegaly, as illustrated in Fig. 1B. Figure 1C illustrates encephalomalacia following traumatic head injury with white matter signal abnormalities, white matter loss, and ventriculomegaly.
Table 2. Contralateral MRI findings between MCD, infarction/IVH, RE, and other
MCD (n = 31)
Infarction/IVH (n = 31)
REa (n = 8)
Otherb (n = 11)
bSturge-Weber syndrome, and encephalomalacia following traumatic head injury, infection, or intrathecal chemotherapy.
cAffecting only one lobe versus more diffuse changes.
dFisher’s exact test [Exact Sig (two-sided)].
White matter loss
Increased CSF space
White signal abnormalities
Indistinct gray–white junction
Cortical signal abnormalities
Cortical thickness abnormalities
Abnormal deep gray
Focalc cortical malformation
By subjective judgment, the abnormal contralateral findings were usually mild (27 patients, 33%) or moderate (41 patients, 51%); in 12 (15%) the findings were severe (Table 2). Figures 1A–D provides examples illustrating each level of severity.
Overall, 80% of patients were seizure-free at latest follow-up 12–84 (median 24) months after surgery. For 10 of the 110 patients, the outcome was assessed after a second surgery because of persistent seizures after functional hemispherectomy. In seven patients a more complete transsection was performed after 2 weeks (2), 5 months (1), 8 months (1), 13 months (1), 3 years (1), and 8 years (1). In three patients an anatomic hemispherectomy was performed after 8 months (1), 13 months (1), and 27 months (1). Shifts in seizure outcome class over time included two patients who stopped having seizures after 6 months, and 11 patients who had recurrence of seizures after 6 months to 5 years.
Seizure outcome did not correlate with presence of contralateral abnormality on preoperative MRI, or with etiology of the lesion (Table 3). Other features that did not correlate with seizure outcome included ipsilateral predominance in ictal or interictal EEG, presence of lateralizing signs during seizures, or diffuse or focal (within one lobe) nature of the contralateral MRI findings. Subgroup numbers were small, but no individual types of contralateral MRI abnormality, listed in Table 2, emerged predictive of poor outcome. Among patients with various types of contralateral MRI abnormality, the percentage of seizure-free patients was 79% overall, ranging from 73% (for children with contralateral focal encephalomalacia) to 81% (for children with contralateral diffuse periventricular leukomalacia). These results did not differ significantly from those in children without contralateral MRI abnormality (83%).
Table 3. Frequency of seizure-free outcome in patients with or without MRI abnormality on the side contralateral to surgery
Abnormal contralateral side
Normal contralateral side
MCD, malformations of cortical development; IVH, intraventricular hemorrhage; RE, Rasmussen’s encephalitis.
aFisher’s exact test [Exact Sig (two-sided)].
When looking at the different subgroups of contralateral abnormality, we found significant correlations with seizure outcome only for cortical signal abnormality and thickness. In the MCD etiology group, the significant correlation was with abnormal cortical signal: 2 (33%) of 6 patients with this finding became seizure-free compared to 20 (80%) of 25 patients without (p = 0.02). In the infarction and hypoxic–ischemia etiology group, the significant correlation was with abnormal cortical thickness: 6 (55%) of 11 patients with this contralateral finding became seizure-free, compared to 58 (83%) of 70 without (p = 0.05). Contralateral white matter changes did not appear to adversely affect seizure outcome; on the contrary, 48 (87%) of 55 patients with this finding became seizure-free, compared to 16 (62%) without (p = 0.02), and these results were similar within each of the etiology subgroups.
Postoperative EEG and seizure outcome
Postoperative EEGs at 6 months were available for retrospective review for 68 of 81 (84%) of children with contralateral MRI abnormality, and for 21 of 29 (72%) of children without contralateral MRI abnormality. Among children with available postoperative EEG, generalized or contralateral epileptiform discharges were present on preoperative EEG in 27 children (40%) with contralateral MRI abnormality, and in five children (24%) without contralateral MRI abnormality (ns). Generalized or contralateral epileptiform discharges disappeared on postoperative EEG in 19 of 27 (70%) of children with contralateral MRI abnormality and in 3 of 5 (60%) of children without contralateral MRI abnormality (ns). Overall, seizure-free outcome occurred in 11 of 19 (58%) of children with generalized or contralateral epileptiform discharges on postoperative EEG, and in 66 of 70 (94%) of children without these postoperative EEG findings (p = <0.001). Among children with generalized or contralateral epileptiform discharges on postoperative EEG, seizure-free outcome occurred for 9 of 16 (56%) with contralateral MRI abnormality and 16 of 18 (89%) without contralateral MRI abnormalities (p = 0.052). Among children without generalized or contralateral epileptiform discharges on postoperative EEG, seizure-free outcome occurred for 50 of 52 (96%) of children with contralateral MRI abnormalities and for 66 of 70 (94%) of children without contralateral MRI abnormality.
Upon retrospective review, we found MRI abnormalities in the contralateral hemisphere, usually mild to moderate in severity, in the majority of children who underwent hemispherectomy for refractory epilepsy due to various etiologies. When additional abnormalities on the contralateral MRI were appreciated preoperatively, the decision to proceed to surgery was based on the more pronounced nature of the lesion on the side consistent with hemiparesis, lateralizing findings on EEG, lateralizing features during seizures, severity of the refractory epilepsy, and low risk of incurring a new postoperative neurologic deficit (Wyllie et al., 2007). The contralateral MRI findings, always less prominent than those in the ipsilateral hemisphere, did not correlate with seizure outcome, and 79% of patients were seizure-free. Our findings indicate that contralateral MRI abnormalities may be the norm in the types of extensive congenital or early acquired hemispheric lesions that occur in the hemispherectomy candidates—especially with use of sensitive modern MRI techniques and critical and careful MRI review—and may not contraindicate hemispherectomy in otherwise favorable candidates.
To our knowledge this is the first systematic follow-up study to assess the impact of contralateral MRI findings on seizure outcome after hemispherectomy for refractory epilepsy due to a variety of etiologies, most often MCD, perinatal infarction, or hypoxic–ischemia. Children with ipsilateral encephalomalacia due to perinatal insults had contralateral white matter changes consistent with global insult, whereas children with MCD had contralateral white and gray matter findings consistent with a bilateral malformative process. The high rate of seizure-free outcome in this series indicates that the epileptogenic zone may be unilateral, even if MRI shows bilateral findings.
We found no correlation between seizure outcome and type of contralateral MRI abnormality—except for contralateral abnormal cortical signal or cortical thickness, each of which suggested a possibly worse outcome in an etiologic subgroup. On the other hand, contralateral white matter changes did not seem to have a great importance for the seizure outcome, since seizure-free outcome was actually significantly more frequent in children with these findings than in those with normal contralateral MRI. Although subgroups were small, our results confirm earlier knowledge that cortical abnormalities in signal and thickness are more strongly associated with epilepsy and epileptogenic lesions than are white matter changes and periventricular leukomalacia (Wyllie et al., 1996; Gurses et al., 1999).
Our results did not demonstrate that severity of the contralateral MRI abnormality correlates with postoperative seizure outcome. However, only a small number of children had contralateral MRI abnormalities that were subjectively judged to be severe, and in these cases the EEG and seizure semiology suggested that the epileptogenic zone was limited to the side with more severe MRI abnormality. Our results cannot be extrapolated to support hemispherectomy in children with severe bilateral abnormalities on MRI in the absence of localizing findings on EEG and seizure semiology. Our study did not include children with bilateral MRI abnormalities who were evaluated for hemispherectomy and turned down for surgery, but only the subset of such patients who were offered surgery because of localizing EEG and seizure semiology.
Some of our patient in both MRI groups had diffuse or contralateral epileptiform discharges on preoperative EEG, consistent with previous reports of successful epilepsy surgery for selected children with early brain lesions and hypsarrhythmia (Delalande et al., 1992; Kossoff et al., 2003; Basheer et al., 2007; Terra-Bustamante et al., 2007) or other types of generalized EEG patterns (Gupta et al., 2007; Wyllie et al., 2007). Generalized or contralateral epileptiform discharges on preoperative EEG tended to be more common among children with contralateral MRI abnormality, but the discharges usually disappeared after surgery among children in both MRI groups. In children with contralateral MRI abnormality, however, the persistence of postoperative generalized or contralateral epileptiform discharges appeared to correlate with persistent seizures, although even in this group the majority of children were seizure-free. Our postsurgical data are consistent with growing clinical experience that in children with refractory epilepsy and early brain lesion, the decision for surgery is no longer based solely on traditional unilateral or focal findings on MRI, clinical semiology, and ictal EEG. Instead it is based on the comprehensive picture of a child with catastrophic epilepsy, mental retardation, and hemiplegia together with unilateral or strongly asymmetric MRI findings (Wyllie et al., 2007).
A limitation of this study is its retrospective nature, but prospective randomized studies of epilepsy for catastrophic epilepsy are not practical. It may be possible with larger patient numbers, and consequently larger abnormality and severity subgroups, to have more significant associations. Long-term follow-up studies will also be important. Also important will be longitudinal studies of developmental outcome after hemispherectomy, and the possible role of contralateral MRI abnormality.
Our results indicate that contralateral abnormalities on MRI should be considered a relative rather than an absolute contraindication to hemispherectomy. In our children selected for surgery, the contralateral abnormalities were always more mild than those on the side targeted for surgery, and almost always mild or moderate rather than severe, but in some cases they were marked. The decision to perform hemispherectomy in such children must be approached with caution and individualized, integrating the whole clinical picture including seizure semiology, EEG, MRI, hemiplegia, severity of the epilepsy, and risk of new postoperative neurologic deficits.
Thank you to Helene Jacobsson, biostatistician, for assistance and advice with the statistics.We confirm that we have read the Journal’s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.
Disclosure: None of the authors has any conflict of interest to disclosure.