Prediction of Seizure-onset Laterality by Using Wada Memory Asymmetries in Pediatric Epilepsy Surgery Candidates

Authors


Address correspondence and reprint requests to Dr. G.P. Lee at Medical College of Georgia, Department of Occupational Therapy (EF-102), Augusta, GA 30912-0700, U.S.A. E-mail: glee@mail.mcg.edu

Abstract

Summary:  Purpose: Because the capacity of intracarotid amobarbital (Wada) memory assessment to predict seizure-onset laterality in children has not been thoroughly investigated, three comprehensive epilepsy surgery centers pooled their data and examined Wada memory asymmetries to predict side of seizure onset in children being considered for epilepsy surgery.

Methods: One hundred fifty-two children with intractable epilepsy underwent Wada testing. Although the type and number of memory stimuli and methods varied at each institution, all children were presented with six to 10 items soon after amobarbital injection. After return to neurologic baseline, recognition memory for the stimuli was assessed. Seizure onset was determined by simultaneous video-EEG recordings of multiple seizures.

Results: In children with unilateral temporal lobe seizures (n = 87), Wada memory asymmetries accurately predicted seizure laterality to a statistically significant degree. Wada memory asymmetries also correctly predicted side of seizure onset in children with extra–temporal lobe seizures (n = 65). Although individual patient prediction accuracy was statistically significant in temporal lobe cases, onset laterality was incorrectly predicted in ≤52% of children with left temporal lobe seizure onset, depending on the methods and asymmetry criterion used. There also were significant differences between Wada prediction accuracy across the three epilepsy centers.

Conclusions: Results suggest that Wada memory assessment is useful in predicting side of seizure onset in many children. However, Wada memory asymmetries should be interpreted more cautiously in children than in adults.

Although the intracarotid amobarbital (Wada) procedure was originally used to assess cerebral language dominance in seizure patients being considered for epilepsy surgery, it was soon modified to include assessment of recent memory to help predict postoperative global amnesia (1,2). Wada memory testing provides a reversible technique for modeling the potential effects of surgery on recent memory functions (3). The association of Wada memory performance and everyday memory function in epilepsy patients has been demonstrated in studies correlating left–right Wada memory disparities with postoperative memory disorders (4,5), hippocampal volume asymmetries determined by magnetic resonance imaging (6,7), and cell densities obtained from pathological specimens of hippocampal tissue after resection (8).

Because demonstrating adequate memory capacity contralateral to the proposed surgical site is thought to be necessary to avoid postoperative amnesia, Wada memory assessment has become a standard part of the preoperative evaluation at many epilepsy surgery centers (9). Unfortunately, Wada testing procedures vary widely among different epilepsy surgery centers, reflecting the training and theoretic biases of the examiners. Such variability may be seen by differences in drug administration, drug dosage, methods of determining drug effect, types of stimuli used, the timing of stimulus presentation, scoring procedures, and memory pass–fail criteria (10,11). Despite differences in Wada methods across epilepsy centers, many investigators have found that Wada memory assessment can provide accurate information about side of seizure onset, especially among adults with mesial temporal lobe seizures (12–15).

When the results of Wada memory testing provide reliable confirmatory evidence of seizure-onset lateralization consistent with noninvasive measures [e.g., EEG, magnetic resonance imaging (MRI), and ictal single-photon emission computed tomography (SPECT)], the necessity for long-term invasive monitoring involving intracranially implanted depth, strip, or grid electrodes may be reduced. Perrine et al. (14) evaluated the classification accuracy of Wada memory asymmetries to predict seizure laterality in 70 adult patients with temporal lobe seizure onset, left-hemisphere language, and no evidence of structural lesions, and they were able to classify accurately 50 (∼71%) of 70 temporal lobe epilepsy (TLE) patients. Of these 50 patients, all of the right, and 96% of the left, temporal lobe focus patients were correctly lateralized by using Wada memory asymmetry scores. Other investigators examined the capacity of Wada memory testing for determining laterality of temporal lobe seizure onset in adult epilepsy surgery candidates and reported correct classification in 80 to 88% of cases (16–20).

Classification of seizure onset laterality has been less successful in non–temporal lobe cases. This is important in pediatric epilepsy surgery patients because mesial temporal lobe seizure onset is less common in children than in adults (21). For example, Spencer et al. (20) found that 66% of patients with neocortical or mesial frontal lobe epilepsy showed Wada memory deficits. These authors concluded that among mesial temporal lobe cases not well lateralized by noninvasive evaluation, and in neocortical or mesial frontal lobe epilepsy, Wada memory testing can provide information regarding localization that may ultimately alter surgical management.

Wada memory asymmetries typically have been less robust in predicting side of seizure onset in children relative to adults. Lee et al. (22) found that the Wada test correctly predicted seizure laterality in 66% of 21 pediatric temporal lobectomy candidates. More recently, Wada memory assessment revealed impaired memory of the epileptogenic side in 59% of 22 pediatric patients with temporal lobe epilepsy (23). Hamer et al. (24) found that the Wada procedure successfully established hemispheric language dominance and memory representation in fewer than two-thirds of 42 preadolescent candidates for epilepsy surgery. Although one investigation (25) reported that Wada memory assessment correctly predicted side of surgery in 91% of children undergoing temporal lobectomy (87% for left temporal lobectomy and 100% for right temporal lobectomy), the memory asymmetry criterion used excluded 42% of children whose memory asymmetry scores were considered indeterminate. Thus, based upon the limited information available, it appears Wada memory asymmetries are more difficult to use for prediction of seizure laterality in children and adolescents than in adults.

Because epilepsy surgery is becoming a more viable treatment option in children, more information is needed about the validity of Wada memory assessment in younger epilepsy populations. Clearly, the risk of developing a postoperative global amnesia is as important in pediatric as in adult epilepsy surgery patients. However, this task is made more difficult because the neurobehavioral relations established in adults may not be applicable to children. Because neural and cognitive systems are in the process of developing in children, cognitive functions (including memory) may have different cerebral representations in children than in adults. Furthermore, the epileptic disease process may have different consequences on the immature, developing brain. More information regarding the capacity of Wada memory assessment to predict lateralization of seizure onset in children is needed. To address this limitation in pediatric epilepsy, three comprehensive epilepsy surgery centers pooled their pediatric Wada memory data and examined the capacity of Wada memory asymmetries to predict side of seizure onset in all children being considered for epilepsy surgery.

METHODS

Subjects

One-hundred fifty-two children (79 boys, 73 girls) age 16 or younger who underwent intracarotid amobarbital (Wada) testing as part of their evaluation for possible epilepsy surgery served as subjects. The children were being evaluated for possible epilepsy surgery at the Medical College of Georgia, Minnesota Epilepsy Group, and Yale University School of Medicine. Mean age was 13.37 years (SD, 2.89; range, 7–16 years). One hundred eight were right-handed, 20 were left-handed, and 24 were mixed-handed. Wada language evaluation revealed that 120 youngsters were left-hemisphere dominant for language, 11 were right-hemisphere language dominant, and 21 showed bilateral cerebral language representation. Ninety-four children had seizures originating from the left hemisphere, and 58 had seizures originating from the right hemisphere. Mean duration of the seizure disorder was 7.10 years (SD, 5.03 years). Seizure onset was determined by 24-h, simultaneous video-EEG monitoring of at least three ictal events. Eighty-seven patients had seizures originating from either the left (n = 56) or right (n = 31) temporal lobes. Sixty-five children had extratemporal seizure foci in the following regions: left frontal lobe (n = 22), right frontal lobe (n = 15), left parietal lobe (n = 9), right parietal lobe (n = 5), left occipital lobe (n = 6), right occipital lobe (n = 4), left frontotemporal region (n = 1), and right frontotemporal region (n = 3). Children with bilateral seizure onset were excluded from the analysis.

Most pediatric surgical candidates underwent Wada testing before surgery at each of the three institutions. At Yale, all surgical candidates underwent Wada testing. At the Medical College of Georgia, all children age 8 and older who were surgical candidates typically underwent Wada testing. Children were excluded if they could not cooperate with baseline tasks because of severe behavior disorders, mental retardation, or muteness. In borderline cases in which it was unclear whether cooperation could be obtained, Wada testing was attempted in children with left hemisphere seizure foci more often than in children with right-sided foci. In Minnesota, children who were very young (younger than 6 years), nonverbal, or profoundly retarded did not always undergo Wada assessment. Children without lateralized seizures who were candidates for anterior two thirds corpus callosotomy typically did not have presurgical Wada testing. All youngsters who were candidates for focal resection underwent the language portion of Wada testing in Minnesota if they were able to comply with the language task demands during pretesting and the memory portion if they understood the task demands for memory items during the pretest. Every youngster in this investigation had both left and right amobarbital injections during Wada assessment.

Procedure

Wada testing was conducted immediately after cerebral angiography. Amobarbital was introduced by hand over a 4- to 5-s interval into the internal carotid artery by using a transfemoral catheter. A single bolus injection was administered in most cases, and incremental injections were administered at two institutions if marked hemiparesis was not induced. At the Medical College of Georgia, 75 mg of sodium amobarbital was typically chosen for children younger than 10 years and 100 mg for children older than 10 years. The Minnesota Epilepsy Group administered 50 mg to children younger than 8 years, 75–100 mg for ages 8 to 11 years, and 100–125 mg to children 12 years or older. Yale administered 75–100 mg to children younger than 10 years, 100 mg to 10–16-year-olds, and 130 mg to patients 16 years and older based on clinical considerations such as size and location of lesion and baseline level of function. Although the mean amobarbital dose differed at each epilepsy center (Medical College of Georgia, 85.14 mg; Minnesota, 105.93 mg; Yale, 109.22 mg), there were no statistically significant differences between left and right hemisphere injections. Mean amobarbital dose injected into the left hemisphere was 100.66 mg (SD, 18.64), and into the right hemisphere, 99.53 mg (SD, 18.43).

Amobarbital was administered to the side of suspected seizure onset first. If seizure laterality was unknown, the right hemisphere was typically injected initially, especially in younger children. All centers performed both left and right hemisphere injections on the same day, and there was a minimum of 30 min between injections at two institutions. At one institution, there was a minimum of 20 min between injections, although 30 min was the approximate modal time between injections.

Wada memory assessment

Memory stimuli were presented within ∼30–60 s after confirmation of contralateral hemiparesis. Although the type and number of memory stimuli (e.g., pictures of familiar objects, line drawings, real objects, geometric shapes, arithmetic problems, printed words, and dolls wearing different outfits of clothing) and methods varied at each institution, all children were presented with six to 10 items. After return to neurologic baseline, as demonstrated by 5/5 strength, normal language, and absence of pronator drift and asterixis, recognition memory for the various stimuli was assessed. At two institutions, recall and recognition of the memory stimuli was initiated a minimum of 10 min after injection. At the Medical College of Georgia, recognition memory was tested for eight real objects that were interspersed with 16 foils. Objects were presented in a randomized sequence, and patients indicated whether each object had been presented earlier in the test. One-half the number of false-positive responses was subtracted from the total number of objects accurately recognized to correct for guessing. The Minnesota Epilepsy Group assessed memory by using a multiple-choice recognition format for most items and, in the adolescent/adult Wada protocol, a Yes/No forced-choice format (with false-positive error correction) for color photograph items. Yale used multiple-choice recognition, consisting of a single card containing the target and five foils, to evaluate recognition memory. Further details of these Wada procedures may be obtained elsewhere (25,26). Because the number of memory stimuli differed at each institution, memory data were calculated as the percentage of items recognized. Group memory asymmetries were analyzed by using a difference score calculated by subtracting the percentage correctly recognized after right injection from the percentage correctly recognized after left injection (left minus right injection). Difference scores ranged from +100 (all items recognized after left injection, and none recognized after right injection) to –100 (all items recognized after right injection, and none recognized after left injection).

RESULTS

Group results

Wada memory asymmetries (i.e., differences in mean memory scores between right and left injections) accurately predicted seizure-onset laterality among the 87 children with unilateral TLE to a statistically significant degree (F1,85 = 20.65; p = 0.0001). As may be seen in Table 1, prediction of seizure-onset laterality was more robust among children with right TLE, probably because of the verbal nature of some memory stimuli being poorly recalled in children with left seizure onset after left (ipsilateral) injection. Nonetheless, as a group, left TLE children recognized significantly fewer stimuli after contralateral injection (t54 = 1.93; p = 0.02).

Table 1.  Percentage of memory stimuli recognized after injection of amobarbital ipsilateral and contralateral to unilateral temporal lobe–onset epilepsy
Side of injectionGroup (N = 87)
 Left TLERight TLE
XSDXSD
  1. TLE, temporal lobe epilepsy.

Ipsilateral injection65.21(29.08)83.55(21.30)
Contralateral injection52.32(32.35)36.32(32.71)

Wada memory asymmetries also predicted seizure-onset laterality accurately in the 65 children with unilateral extratemporal lobe (ExTL) seizure onset to a statistically significant degree (F1,85 = 7.63; p = 0.0009). As in children with unilateral TLE, prediction of seizure onset laterality was more robust among children with right ExTL seizures. However, as a group, there was no significant difference in the percentage of memory stimuli recognized after left- and right-hemisphere injection among left ExTL seizure-onset children (t40 = 1.15; p = 0.13; Table 2).

Table 2.  Percentage of memory stimuli recognized after injection of amobarbital ipsilateral and contralateral to extra–temporal lobe epilepsy
Side of injectionGroup (N = 65)
 Left ExTLERight ExTLE
XSDXSD
  1. ExTLE, extra–temporal lobe epilepsy.

Ipsilateral injection58.94(32.52)70.84(32.49)
Contralateral injection51.55(34.12)40.00(35.05)

Individual patient prediction

Although Wada memory asymmetries accurately predicted side of seizure onset by using group data, prediction among individual cases was less encouraging. Children with left TLE were frequently misclassified regardless of which classification rule was used. When using a classification criterion of a ≥20% difference between left- and right-injection memory scores, left TLE children were incorrectly classified in 27% of cases. In contrast, seizure-onset laterality in children with right TLE was never incorrectly predicted by using the ≥20% difference criterion (χ2 = 11.77; df = 1; p < 0.01). Use of the 20% difference criteria resulted in 31 (36%) of 87 children not being assigned to either the left or right TLE groups. These results are given in Table 3.

Table 3.  Prediction of seizure-onset laterality among individual cases of temporal lobe epilepsya
Classification accuracyGroup
Left TLERight TLE
  1. Criterion: ≥20% difference between left and right memory scores.

  2. TLE, temporal lobe epilepsy.

Correct prediction19/55 (35%)22/32 (69%)
Incorrect prediction15/55 (27%)0/32 (0%)
Not lateralized (memory score
 difference <20%)
21/55 (38%)10/32 (31%)

Similar statistically significant results were obtained in temporal lobe cases when using a classification criterion of a memory recognition score asymmetry (of any magnitude) in the correct direction (viz., memory scores after injection ipsilateral to seizure focus greater than memory scores after contralateral injection). Ties were considered incorrect predictions. Although fewer left-onset children were misclassified by using this criterion (with the 20% memory score difference criterion, 65% of children were either misclassified or unable to be classified), a majority of cases were nevertheless incorrectly predicted (left TLE, incorrect prediction = 52%). More right-TLE children were misclassified with this classification criterion (right TLE, incorrect prediction = 16%) than the 20% difference criterion (right TLE, no incorrect predictions). Despite the disparity between classification criteria, the classification rule of a memory recognition score asymmetry (of any magnitude) in the correct direction was nevertheless able to predict accurately seizure-onset laterality among children with temporal lobe seizures to a statistically significant degree (χ2 = 28.88; df = 1; p = 0.0001; Table 4).

Table 4.  Prediction of seizure-onset laterality among individual cases of temporal lobe epilepsy
Classification accuracyGroup
Left TLERight TLE
  1. Criterion: memory scores after injection ipsilateral to seizure focus greater than memory scores after contralateral injection.

  2. TLE, temporal lobe epilepsy.

Correct prediction26/55 (48%)27/32 (84%)
Incorrect prediction29/55 (52%)5/32 (16%)

There were no statistically significant differences between prediction of side of seizure onset in youngsters with left or right ExTLE when using a classification criterion of a ≥20% difference between left and right injection memory scores. Examination of Table 5 reveals that only 37% of left, and 63% of right, ExTLEs were correctly classified by using this Wada memory score disparity of ≥20%. This memory score criterion was unable to shed any information in 16 (25%) of the 65 ExTLE cases.

Table 5.  Prediction of seizure-onset laterality among individual cases of extra–temporal lobe epilepsy
Classification accuracyGroup
Left ExTLERight ExTLE
  1. Criterion: ≥20% difference between left and right memory scores.

  2. ExTLE, extra–temporal lobe epilepsy.

Correct prediction13/35 (37%)19/30 (63%)
Incorrect prediction11/35 (31.5%)6/30 (20%)
Not lateralized (memory score
 difference <20%)
11/35 (31.5%)5/30 (17%)

There was no statistically significant difference between correct and incorrect predictions based on memory score asymmetries in children with left or right ExTL seizure onset by using the criterion of any memory score asymmetry in the correct direction (χ2 = 1.35; df = 1; p > 0.05; Table 6). Surprisingly, Wada memory asymmetries correctly predicted side of seizure onset among left ExTL children with similar accuracy to that seen in left TLE (left ExTL, correct prediction = 53%; left TLE, correct prediction = 48%). Unfortunately, however, using the criterion of any memory asymmetry in the correct direction resulted in less accurate seizure laterality prediction among right seizure-onset children (right ExTL, correct prediction = 67%; right TLE, correct prediction = 84%; Tables 4 and 6).

Table 6.  Prediction of seizure-onset laterality among individual cases of extra–temporal lobe epilepsy
Classification accuracyGroup
Left ExTLERight ExTLE
  1. Criterion: memory scores after injection ipsilateral to seizure focus greater than memory scores after contralateral injection.

  2. ExTLE, extra–temporal lobe epilepsy.

Correct prediction20/38 (53%)18/27 (67%)
Incorrect prediction18/38 (47%)9/27 (33%)

Age effects

Because language functions are in the process of development in children, differences in Wada language and memory results between adults and children may be due to incomplete cerebral localization and lateralization of these functions among younger children (27). Linguistic functions may facilitate the encoding of new memories regardless of the type of material to be recalled. To determine if age was a factor in the relative incapacity of Wada memory asymmetries to predict side of seizure onset in children with left TLE, age effects were examined in children with left TLE who were right-handed and left hemisphere dominant for language (n = 54) by using the criterion of any memory score asymmetry in the correct direction. Wada memory asymmetries were in the predicted direction (memory after language dominant, left injection memory better than memory after nondominant right injection) in 26 (48%) of 54 left TLEs and incorrectly predicted in 28 (52%) of 54 left TLE children. Left TLE children whose side of seizure onset was incorrectly predicted were significantly younger (mean, 11.5 years) than left TLE children whose seizure-onset laterality was correctly predicted (mean, 14.1 years; t = 3.13; df = 52; p = 0.001).

DISCUSSION

The overall capacity of Wada memory assessment to classify seizure onset laterality accurately in children in this study was 69%. This is roughly equivalent to similar investigations that have demonstrated impaired Wada memory scores in 50 to 65% of children (22–24), but lower than correct classification rates reported in studies of adult epilepsy surgery candidates, which have ranged between 70 and 88%(14,16–20). Wada memory asymmetries in the current study also predicted the side of seizure onset better in children with right hemisphere seizure foci than in children with left hemisphere seizure onset. Thus our results, in conjunction with previous studies (22,23,28–30), indicate that Wada memory asymmetries should be interpreted more cautiously among children than in adults, especially among children with left hemisphere seizure foci.

Results of the present study also suggested that Wada memory asymmetries are most useful in predicting side of seizure onset among children with TL seizures, although they also are marginally helpful in determining side of seizure onset in the majority (59%) of ExTL-onset cases. This somewhat surprising finding has been reported in prior investigations as well. For example, Westerveld et al. (25) obtained lateralizing Wada memory results in 52.4% of children with TL seizure foci and in 50% of children with seizures originating outside the TLs. Spencer et al. (20) reported that Wada memory asymmetries provided useful information regarding localization that ultimately altered surgical management in 66% of patients with neocortical or mesial frontal lobe epilepsy. These findings are most likely due to the contribution of other cognitive functions to recent memory. Although the hippocampus, lying deep within the mesial temporal lobes, is clearly vital for the formation of new memories, disruption of other cognitive functions mediated by extratemporal brain regions also may negatively affect new learning.

Previous investigations have yielded conflicting results as to whether Wada memory results are less valid among younger children. Neither Westerveld et al. (25) nor Hempel et al. (30) found a difference in overall Wada memory performance as a function of age. In contrast, other investigators (28,29) reported that Wada memory scores were lower in younger relative to older epilepsy patients. These conflicting results are most likely due to sample and methodologic differences between the studies, most notably, variability in age range, type and location of seizure onset, drug administration and dosage, type of stimuli used, and the Wada memory pass–fail criteria used. For example, if children with left hemisphere seizure onset were considered for epilepsy surgery at a younger age than children with right hemisphere seizures, having an unequal representation of left hemisphere seizure-onset patients would skew the results because left seizure patients perform worse as a group on Wada memory testing.

Even though the epilepsy surgery sample size in the current study was larger than that in most investigations, there were almost twice as many left (n = 94) than right (n = 58) seizure-onset children in our sample. To correct for this unequal representation of left versus right seizure onset, age effects were examined only in children with left TLE who were right-handed and left hemisphere dominant for language. Left TLE children whose side of seizure onset was incorrectly predicted were significantly younger than left TLE children whose seizure-onset laterality was correctly predicted. This suggests that a younger age may hamper the ability of Wada memory asymmetries to predict side of seizure onset in children with left TLE.

There may be several reasons for the unusually high proportion of left seizure foci in the current sample. The Medical College of Georgia's selection criterion was biased toward having more children with left, than right, seizure onset undergo Wada testing. This selection bias was present because many physicians thought the potential negative consequences of left focal resections are more detrimental to the child's overall postoperative development. Another possibility may be that more children with left seizure foci are referred for consideration of seizure surgery at an earlier age because the adverse effects of left-sided seizures on cognitive development (particularly language) and educational performance are more salient. The unusually high percentage of right or bilateral language representation in this sample (21%) relative to adult samples also may be related to the large proportion of left-sided cases. Early left-hemisphere disease most likely increased the likelihood of having anomalous language representation in this sample of youngsters. Thus because there are so many left-sided cases in this sample, early damage may have caused language to reorganize into healthy brain regions such as the neurologically intact right hemisphere.

Finally, there were large differences in Wada prediction accuracy among the three Epilepsy Surgery Centers. Anecdotal examination of the memory stimuli and scoring criteria used at each center suggested that reliance on verbal stimuli to determine Wada memory asymmetries may result in poor prediction of seizure-onset laterality in children with left-hemisphere seizure foci. Wada methods and memory stimuli type may be critical in determining its success in predicting seizure-onset laterality in children.

Ancillary