Long-Term Functional Outcome Following Left Hemispherotomy in Adults and Pediatric Participants with Fmri Analysis
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/0028-3886.355100
Source of Support: None, Conflict of Interest: None
Keywords: Epilepsy, fMRI, hemispherotomy, language, plasticity
Hemispherotomy is largely reserved for the pediatric age group as the brain plasticity favors good functional outcomes. There are even reports in the literature with age limits to guide as to when the surgery can be offered and what functional outcome is to be expected in a pediatric population. However, there is a dilemma for surgery in adults who present with drug-refractory “dominant” lobe (left) hemispheric epilepsy but are ambulant and eloquent preoperatively. This is due to the fact that the preoperative magnetic resonance imaging (MRI) of such a patient usually shows some areas of normal cortical mantle amidst the largely encephalomalacic left hemisphere, which are usually not the areas maintaining the motor and language functions. Though it is a known fact from the pediatric studies that the right hemispheric functional transfer is responsible to maintain cognitive functions, this factual knowledge alone is not good enough to proceed with hemispherotomy in adults. The critical balance between seizure freedom and a good functional outcome is of paramount importance in such a scenario. An attempt was made to aid the clinical judgment with objective preoperative clinical evidence of right hemispheric transfer with a comparison to postoperative functional outcomes in adults. There are case reports and a few case series discussing mostly the surgical safety and a few of them discussing the motor functional outcome after hemispherotomy in adults., However, neither there are studies where comparisons have been drawn with the pediatric age group in the same center nor there has been a usage of functional MRI (fMRI) as a preoperative tool for the adult group. Moreover, there could be bilateral language representation in case of incomplete perinatal infarcts, which again could lead to poor postoperative outcome. Hence, the objective is to find the subset of adult population in whom the hemispheric surgery can be offered with reasonably good functional outcome and to know the role of a noninvasive fMRI investigation in decision-making.
A retrospective analysis of data was done on a homogenous group of 20 left hemispheric epilepsy participants who underwent endoscope-assisted interhemispheric vertical hemispherotomy performed by a single surgeon over 5 years (2013–2018) at the All India Institute of Medical Sciences (AIIMS, New Delhi). Institutional Ethics Committee approved the study. The preoperative and postoperative clinical data of five adults and 13 pediatric participants (n = 18; attrition of participants- two) were collected. This included along with demographic profile, seizure data, detailed neurological evaluation of language, neuropsychological assessment, and the anatomical and fMRI radiological data.
Participants underwent complete language evaluation preoperatively and postoperatively with parameters of word recognition, phoneme discrimination, picture–word matching, auditory comprehension, paragraph comprehension, picture naming, and repetition with standardized norms. Handedness was determined by the Edinburgh inventory.
Muscle strength of the affected upper and lower extremities was assessed according to the criteria for manual muscle testing, using a 6-point scale (Medical Research Council (MRC) range 0–5).
The Indian Aphasia Battery (Nehra et al., 2013) scale was administered to evaluate language functions. Aphasia rating was analyzed based on language error rates; subtests/category of the IAB, including speech-language dysfunction (SLD) and visual reading dysfunction (VRD), had scores of more than 20 (20%) out of 103 and 12 (27%) out of 44, respectively. Based on these tests, the language dysfunction was classified into good, poor, and telegraphic. In telegraphic language, the comprehension is intact with poor performance in naming and repetition.
fMRI study was carried out using 32-channel head coils on a 3.0 T MR scanner (Ingenia 3.0 T; M/s. Philips, Texas, USA) in the Department of Nuclear Magnetic Resonance (NMR) (AIIMS), with standard parameters.
Language fMRI task
Language task design and standardization details for semantic and syntactic processing have been described in literature. Briefly, the language task consisted of semantic word generation and syntax comprehension components. The task was visually cued through SuperLab presentation software (Cedrus Inc., San Pedro, CA, USA). Subjects were instructed to read simple semantic words during the word generation task. During the comprehension syntactic–semantic task, subjects were requested to correct the syntax of sentences and read out. Each patient underwent an fMRI scan twice, one preoperatively and the other 6 months postsurgery. Before entering the MRI, the subjects were instructed and trained to execute the tasks by using visual cues.
Motor fMRI task
The motor task was incorporated into a block design with five cycles of activity alternating with the resting baseline. The active motor task was employed for the right upper/lower limbs in healthy controls and the passive motor task for the study participants. The passive motor paradigm was used for participants with hemiparesis with power of grade <3. In the active motor task, healthy controls were instructed to perform flexion and extension movement of the elbow. During the passive task, patient's wrist, fingers, ankle, and toes were moved by the investigator using flexion–extension.
Statistical fMRI analysis
The fMRI data for language and motor tasks were analyzed by Statistical Parametric Mapping (SPM) software (Wellcome trust, London, UK; version SPM 8) using MATLAB (MathWorks Inc., Natick, MA, USA). The data were preprocessed using slice timing, motion correction, co-registration, normalization, and smoothing steps.
The laterality index (LI) was computed to evaluate the laterality/dominance of frontal and temporal lobe regions in the healthy controls and participants. The regional LI was measured using the formula [LI = (left − right)/(left + right)], resulting in positive values for predominantly left hemisphere lateralization (+1) and negative values for right hemisphere lateralization (−1), which has been reported earlier.
Correlation analysis (fMRI LI and language outcome)
The Pearson correlations (using Statistical Package for the Social Sciences [SPSS] version 24) were tested between LIs (from language: word generation and syntax comprehension tasks) and the functional language outcome with the α level set at P < 0.05. A strong negative and positive correlation was interpreted in the range of −1 to 1.
The technique of endoscope-assisted hemispherotomy has already been described in detail previously.,
Clinical and demographical results: Of the 20 participants, based on the age at presentation for surgery, the study population (n = 18) was divided into two groups of five adults (mean age: 20.5 years, range: 19–23 years) and 13 pediatric patients (mean age: 9.2 years, range: 1.5–15 years). Two of the 13 pediatric participants had a global developmental delay at presentation and remained the same postoperatively. The patient characteristics of the remaining 16 participants are shown in [Table 1] and [Table 2]. All of them were operated on for drug-refractory hemispheric epilepsy and had International League Against Epilepsy (ILAE) class I seizure outcome. Complications and/or adverse events occurred in 10 participants, with the most common being postoperative fever (n = 9). They had no evidence of infection, and the cause was considered aseptic meningitis due to blood in the ventricles and in most, fever subsided within 1 week. Additional complications included septic meningitis (n = 1) and hyperammonemia (n = 1). Both of the latter two participants were treated appropriately and improved completely. Further results and discussion will be focused on the stated objective of analyzing the language and motor functional outcome.
Age at onset of weakness: Based on the extent of weakness and the age at onset, the pediatric participants were divided into three groups, as the outcomes were different in each of these as described in the following sections. Group I participants (n = 4) had significant weakness in terms of hemiparesis below the age of 7 years (mean age of 1.9 years at the onset of weakness), and two of them (2/4) had in addition delayed milestones (n = 2) at presentation. In group II (n = 2) participants, the onset of hemiparesis was noticed at an age above 7 years (mean: 10 years) and group III participants (n = 5) had only distal weakness at presentation that was noticed at the mean age of 3.7 years (range: 2.5–5 years). Among adults, four had hemiparesis with the onset seen at the age of <1 year, and in one, the weakness was noted at the age of 14 years.
Language functional outcome: Of the 13 pediatric participants, two had global developmental delay. Group I (n = 4) participants had the best outcome. Of the four participants, two (participants 6 and 7) presented with hemiparesis in addition to delay in other milestones at the age of 1.5 and 4 years, respectively. The etiologies were porencephaly and hemimegalencephaly (HM), respectively. Following surgery performed at the presentation, they eventually acquired language and motor milestones at a mean follow-up period of 55 months. The language was not normal, but comprehension and fluency parameters of the testing were satisfactory with the ability to frame meaningful sentences. Other two participants (participants 8 and 9) presented with hemiparesis at the age of 6 months and 5 years with a diagnosis of hemispheric cortical dysplasia (HCD) and Rasmussen's encephalitis (RE), respectively. Both participants presented with hemiparesis and normal milestones achieved before surgery, which was performed at the age of 13 and 6 years, respectively. Both had impaired comprehension and verbal fluency preoperatively. Though the language function deteriorated initially, at a mean follow-up of 32 months, both comprehension and fluency parameters improved significantly. Group II participants (n = 2) presented with hemiparesis with weakness noticed at 9 and 11 years (participants 15 and 16), respectively, due to RE. Preoperatively, the language function tests revealed intact comprehension for simple commands and significantly decreased word output (telegraphic language). Following the surgery at a mean follow-up period of 24 months, the language function remained same with no further improvement. Majority of the participants belonged to group III (n = 5) (participants 11–14) with etiologies of HCD, RE, HM, and post-encephalitic gliosis. All of them presented only with minimal distal weakness at the wrist and ankle, with the mean age at onset being 3.5 years. Their language was telegraphic at presentation, and at a mean follow-up period of 42 months, this remained the same.
Of five adults, four (participants 1–4) presented with hemiparesis noticed during infancy, and they had intact language function parameters at presentation. The etiology was post-infarct hemispheric gliosis in all of them. All of them developed seizures between the ages of 3 months and 2 years. The surgery was performed in view of drug refractory epilepsy (DRE) with clear lateralization to the left. Four participants retained normal comprehension and verbal fluency components of the language postoperatively at a mean follow-up period of 38 months. The remaining one patient (n = 1/5, patient no. 5) had Broca's aphasia and hemiparesis at presentation and developed seizures at the age of 18 years due to RE. She clinically presented and was operated on at the age of 22 years in view of uncontrolled seizures. Preoperatively, her language function was telegraphic with good comprehension and ability to speak a few comprehensible words, but with poor results in other parameters. Postoperatively, the language function deteriorated further, affecting verbal fluency more than comprehension. She was able to comprehend simple verbal commands. However, she was able to make only incomprehensible sounds on reading tasks and other language tests at a follow-up period of 24 months.
Motor functional outcome: Of the 13 pediatric participants, group I patients achieved normal milestones postoperatively (n = 2) as described earlier and were ambulant independently. Two others were ambulant preoperatively with right hemiparesis. In these participants, the upper limbs (grade 3 proximally and grade 1 distally) were more affected than lower limbs (grade 4 + proximally and grade 3 distally). Postoperatively, they had mild deterioration of power, but at follow-up (mean follow-up of 32 months), both regained preoperative power and were ambulant independently. Group II participants presented with hemiparesis and group III with only mild distal weakness in the wrist and ankle. Both the groups deteriorated further postoperatively, but they recovered to an ambulatory state at a mean follow-up period of 36 months. All participants had significant weakness of hand preoperatively and postoperatively, with no handgrip seen even in a patient with a maximum follow-up period of 72 months.
Of the five adult participants who had right hemiparesis at birth but were ambulant preoperatively, only one patient became wheelchair bound postoperatively at a mean follow-up period of 30 months. This is the same adult with RE who, in addition to language function deterioration, developed worsening of right hemiparesis from grade 4 to grade 1 power in lower limbs and from grade 3 to grade 2 in upper limbs. The remaining four participants, who had post-infarct encephalomalacia as an etiology, regained their preoperative power at 38 months follow-up and were ambulant independently. The hand function remained poor both pre- and postoperatively, similar to the pediatric age group participants. The clinical results have been shown in a flow diagram [Figure 1].
Neuropsychology: In adults, the mean intelligence quotient (IQ) level was 87, indicating a low average intelligence level. The immediate, delayed, and long-term recall components of verbal memory were impaired. The visual memory was mostly intact, but with impairment in visuo-constructive ability. The sustained attention was intact, but the mental flexibility and set-shifting tests were impaired. In the pediatric group of participants, the results were almost similar but with lower IQ range than adult subsets.
Task-based fMRI results
Language task fMRI: In the group of adult participants, all four who retained normal language postoperatively showed activation of right inferior frontal gyrus (IFG) and middle frontal gyrus (MFG) during the semantic word generation tasks [Figure 2]a with no activation seen in left hemisphere [Figure 2]b. The activation was seen in the right superior temporal gyrus (STG) and middle temporal gyrus (MTG) during the syntax comprehension tasks [Figure 2]a, which were significantly different from the controls [Figure 2]g, [Figure 2]h, [Figure 2]i, [Figure 2]j. Again, there was no significant change noted in the postoperative scans. The remaining one patient who deteriorated postoperatively had shown activation in the left hemisphere preoperatively in locations similar to controls, as explained earlier, and the test could not be done postoperatively due to deterioration in language function.
Of the 13 pediatric participants, only eight could complete the language tasks. All the eight subjects complied with the instructions, and language tasks were carried out preoperatively in all and postoperatively in some (at a follow-up period of 6 months). The results were compared with those of six healthy controls. In two (n = 2/4) participants of group I with normal language function, during semantic word generation, the blood oxygen level dependent (BOLD) activation was observed in the right IFG and MFG [Figure 2]a. However, the location was more posterior close to the sensory–motor cortex, compared to the controls [Figure 2]g, [Figure 2]h, [Figure 2]i, [Figure 2]j. During the syntax comprehension task, BOLD activation was observed in the right STG and MTG both pre- and postoperatively, compared to controls [Figure 2]a. Of the seven participants with telegraphic language, six (group II [n = 2] and group III [n = 4]) could complete the tasks preoperatively sufficient enough for analysis. In this group, during the syntax comprehension task, BOLD activation was observed bilaterally in the STG, MTG, and postcentral gyri [Figure 2]e and [Figure 2]f. However, the extent of activation was more on the right compared to the left. Similarly, during the semantic word generation task, the activation was seen bilaterally in the IFG and MFG. But the activation was more in the left hemisphere, as shown in [Figure 2]c and [Figure 2]d. This was congruous with the clinical outcome of better comprehension compared to word generation. Both pre- and postoperative results were similar without any change in the location of activations.
LI-scatter plot [Figure 4]: As expected, a significant difference was observed between good and poor language lateralization for frontal regions (t[df = 14] = 5.71, P < 0.001) and temporal regions (t[df = 14] = 5.79, P < 0.001), estimated by independent sample t-test (SPSS 24). The data suggest that an increasing trend toward right hemispheric regional laterality was associated with good language outcome and the left hemispheric laterality was poor. Results indicate that the right hemispheric reorganization was complete preoperatively in participants who had acquired the weakness before the age of 7 years and the pattern remained the same postoperatively.
Motor fMRI: During the active and passive flexion paradigms, BOLD activation showed two different patterns preoperatively. In the first type, which was noticed only in two adults who were ambulant independently, the activation was seen in the ipsilateral precentral and postcentral gyri [Figure 3]a. This area was broader and located more posterolaterally, compared to the negligible ipsilateral motor area activation seen in healthy controls [Figure 3]c, [Figure 3]d, [Figure 3]e, [Figure 3]f. In the remaining 13 pediatric age group and three adult participants, the activation was similar to healthy controls, with activation observed in the contralateral sensory–motor gyri. There was no change in the activation postoperatively.
One always ponders about the correct surgical decision when an adult who is ambulant with intact language presents with left hemispheric DRE. Does the word “dominant” have relevance in these participants when the left hemisphere is clearly diseased? The analysis has been done to answer this by comparing the clinical outcome of adult and pediatric hemispherotomy participants with the fMRI findings. The findings suggest that the extent of weakness and the age at onset of weakness are the most important determining factors for the long-term language outcome.
Age at onset of weakness and language dominance: Age at onset of weakness has been the most important predictor of the language outcome in this study, irrespective of the age at surgery. This is the result of handedness influencing lateralization of the language, the knowledge of which is not new, and it has been applied to study participants to decide in favor of hemispheric disconnection of the left hemisphere in adults and the long-term results have been described. However, in practice, it was not easy to predict which of the participants would fit into this good outcome category. The sole indication for surgery in a complex scenario was to achieve seizure freedom when other measures had failed. In these situations, both developments of dominance (in participants 6 and 7) and plasticity (in participants 8 and 9) have played roles to achieve good functional outcomes. The fMRI findings suggested three different patterns in accordance with the literature and were comparable with clinical outcome. Firstly, all the participants with pathological left-handedness (though they are difficult to differentiate from natural left-handers) had a right hemispheric shift in the language area fMRI activations. Second pattern of bilateral activations was in participants with telegraphic language, and the third pattern was with left hemispheric activations in right-handed participants. It was clear from these findings that early extensive left hemispheric lesions (<7 years) rendered the left hemisphere functionally nondominant and led to the development of right hemispheric language dominance and left-handedness. Historically, this has been demonstrated partially in pediatric age group using Wada test. More recent studies have used fMRI as a tool to study the plasticity following early left hemispheric lesions. This noninvasive tool has replaced the invasive Wada test for language studies, with relevance left for few specific indications. Lidzba et al. looked at the time window for the rightward transfer using fMRI and concluded that the language outcomes were good for those acquiring lesions below 5 years. However, late plasticity has also been reported beyond the age of 6 years, which was similarly found in one of the participants (participant no. 4). Liegeois et al. presented their series using fMRI on six hemispherectomy participants in the pediatric age group, three of whom were left sided, and suggested that the right homologs of Broca's area subserved the language functions. A study similar to all these aforementioned participant series also demonstrated the syntactic capabilities of isolated right hemisphere following left hemispherectomies done in early childhood. There are few participant reports and participant series on adult hemispheric surgeries, and some of these have reported the functional language outcome. However, none of these have correlated the clinical outcome with fMRI findings, nor have they used it preoperatively for the selection of candidates. The largest series by McGovern et al. had 47 participants, of whom 17 had undergone left hemispherectomy. The majority did not undergo any language localization preoperatively, except for a few (n = 7) in whom Wada testing was done. The prognosis was apparently good with only one deteriorating postoperatively. However, since fMRI was not included in the study, it was hypothesized that the prognosis was good if the age of insult was below 6 years and extensive enough to involve perisylvian regions. Similarly, Cukiert et al. considered the affected hemisphere functionally nondominant based on the perioperative clinical cognitive status without any objective evidence. There are instances where such an assumption and even preoperative Wada investigation have failed. Loddenkemper et al. reported a participant of adult hemispheric surgery in whom the insult was acquired at the age of 5 years but developed expressive aphasia postoperatively. This was the case of unexpected bilateral language dominance, which had passed the Wada test preoperatively. The fMRI would have shown the bilateral representation, thereby overcoming this disadvantage of the Wada test. Similarly, Liang et al. reported impairment of language function in 7/15 participants who underwent left hemispherectomy. This was thought to be due to reduced plasticity, though in four of them, the age at onset of weakness was below 7 years. Again, no fMRI correlation was included in the study. The neurophysiological basis of pathological handedness and lateralization of language dominance has been supported and implicated both by the evolutionary and functional studies that looked at the development of healthy volunteers' natural handedness. This was thought to be due to the synchronization of vocal, facial, and hand gestures. This probably would also explain the relation between the extents of weakness determining the functional transfer of language.
Motor functional outcome: From this study, it was clear that the perinatal insult and those weaknesses with onset below 7 years had a good prognosis. But the prognosis was good only in those who had holohemispheric insult as in infarcts, which produced significant hemiparesis (as opposed to those who had only distal weakness) (patient nos 1-4 and 6–9). Similar findings were noted in other studies as well. Also, similar to other series, the postoperative deterioration improved to an extent of independent ambulation on long-term follow-up, albeit with a paretic hand and residual ankle weakness. The fMRI has shown the role of ipsilateral motor cortex reorganization in participants with a good outcome., The passive paradigms employed in this study have been proven equivalent to the active motor tasks., The ipsilateral reorganization has been reported to follow regionalization, with activations seen in the association area rather than in the primary motor cortex. Though this played a significant role in recovery, it is not a robust mechanism as the incidence of recovery is not uniform across studies. Hence, the outcome has multifactorial influence in addition to cortical reorganization like ipsilateral uncrossed anterior corticospinal tract and axonal sprouting. These have been shown in studies using rodent and cat models, and have been further validated by diffusion tensor imaging (DTI) studies.
Etiology: The etiology of complete middle cerebral artery (MCA) infarct acquired in the perinatal period played a significant role in good functional outcomes in the study participants, similar to other studies., In RE, the timing and the extent of weakness play significant roles in determining the postoperative outcome. Participants in this study who had onset below 7 years fared better than those above this limit. However, studies have reported different age limits starting from 5 years up to 9 years.
The major limitation was the smaller number of adult participants. However, the rarity of hemispheric epilepsy in adults and the need to make the group homogenous with a single side accounted for the same. This has been overcome by comparing the pediatric group from the same center and the use of functional imaging. In addition, the application of DTI could have complimented this study well.
Hemispherotomy can be safely advised in adults with good language and functional motor outcome, provided the age at onset of hemiparesis (as opposed to distal weakness) is below 7 years. The results were better in participants with the etiology of perinatal MCA infarct. The fMRI is a very useful noninvasive tool to demonstrate the transfer of functions, plasticity, or reorganization, and it is strongly recommended to use it preoperatively to prognosticate the outcome.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
[Figure 1], [Figure 2], [Figure 3], [Figure 4]
[Table 1], [Table 2]