- Top of page
- MATERIALS AND METHODS
Summary: Purpose: Functional mapping of eloquent cortex with electrical neurostimulation is used both intra- and extraoperatively to tailor resections. In pediatric patients, however, functional mapping studies frequently fail to localize language. Wada testing has also been reported to be less sensitive in children.
Methods: Thirty children (4.7 – 14.9 years) and 18 adult controls (18–59 years) who underwent extraoperative language mapping via implanted subdural electrodes at the NYU Comprehensive Epilepsy Center were included in the study. Ten children and 14 adults underwent preoperative Wada testing. Success of the procedures was defined as the identification of at least one language site by neurostimulation mapping and determination of hemispheric language dominance on the Wada test.
Results: In children younger than 10.2 years, cortical stimulation identified language cortex at a lower rate than was seen in children older than 10.2 years and in adults (p < 0.05). This threshold, demonstrated by survival and χ2 analysis, was sharply defined in our data set. Additionally, Wada testing was more likely to be successful than was extraoperative mapping in this younger age group (p < 0.05).
Conclusions: Analysis of our series demonstrates that language cortex is less likely to be identified in children younger than 10 years, suggesting that alternatives to the current methods of cortical electrical stimulation, particularly the use of preoperative language lateralization, may be required in this age group.
A crucial function of a comprehensive preoperative evaluation for surgery for medically refractory epilepsy is minimizing the risk of postoperative language, cognitive, and sensorimotor deficits. Neuropsychological assessment, the intracarotid sodium amobarbital (Wada) test, and extra- and intraoperative cortical stimulation are commonly used to reduce these potential morbidities. Cortical stimulation allows the most precise localization of eloquent cortex (Bauman et al., 2005) and appears to be safe in both adult and pediatric populations (Wyllie et al., 1988; Adelson et al., 1995).
Although the sensitivity of cortical stimulation in adult epilepsy and tumor surgery series is established, its utility in pediatric populations is less consistent. Functional mapping has identified motor cortex in children as young as 3 (Signorelli et al., 2004) and 4 years (Chitoku et al., 2001), and language cortex in children as young as 4 years in a nontumoral epilepsy case (Chitoku et al., 2001) and 2 years in a tumoral patient (Duchowny, 2001). However, functional mapping for language localization often produces more unpredictable results in children based on preoperative neurologic and neuroradiologic evaluation (Berger et al., 1989). Factors that complicate functional mapping with extraoperative cortical stimulation include incomplete functional maturation of language networks and, in some cases, abnormal functional organization related to the epilepsy etiology. For instance, cortical malformations can affect the distribution of the motor cortex (Akai et al., 2002) and afterdischarge thresholds (Chitoku et al., 2003). Lesions acquired before the age of 5 years may result in displacement of language sites (Duchowny et al., 1996). The level of patient cooperation may also be a factor in the ability of a mapping procedure to identify language function.
A paucity of data exist concerning success rates of language localization in pediatric series (Berger et al., 1989; Ojemann et al., 2003). We lack even survey data on the age at which the ability of extraoperative neurostimulation mapping can reliably identify language sites. The purpose of the current report is to identify success rates in functional mapping for language localization in a pediatric sample and, more specifically, to determine the age at which such a procedure reaches the adult level of reliability, as measured by the ability of the procedure to detect at least one language site.
- Top of page
- MATERIALS AND METHODS
A total of 15 boys and 15 girls (4.7–14.9 years of age; mean, 9.9) underwent functional cortical mapping of language function with implanted subdural grid electrodes. One boy (age 13) underwent two procedures separated by 6 months. Ten of the 30 subjects underwent preoperative Wada testing for lateralization of language function. Additionally, 11 children (ages 10.1–14.8) underwent Wada testing without a subsequent language-mapping study.
In each case, language mapping was performed because the seizure-onset zone was near suspected language areas. In 16 of the 31 mappings, at least one language site was identified. Children younger than 10.2 years had negative mappings at a rate that was significantly higher than that in both older children and adults. The median age of patients with negative (7.8; 95% CI, 6.5–9.3) as opposed to positive (13.3; 95% CI, 10.8–13.7) language mappings was significantly different (Mann–Whitney two-tailed test, p < 0.05). Survival analysis demonstrated a sharp increase in the success rate between 8 and 12 years of age. By dividing the subjects into two groups based on age and selecting the division yielding the maximal χ2 value, we determined that the sharpest increase in positive language mappings occurs at age 10.2 in our data set. Similarly, when comparing children older than 10.2 years with adults, no significant difference was found in the incidence of positive language findings, whereas children younger than this age demonstrated a significantly lower rate of language findings compared with adults (p < 0.05).
Ten children in the study underwent both extraoperative neurostimulation language mapping and preoperative Wada testing, of whom six, or 60%, were successfully lateralized with the Wada test. Language lateralization by both modalities was concordant in five of the six cases, whereas in the sixth case (patient 16), extraoperative stimulation failed to identify language (Table 1). When the 11 subjects who underwent Wada testing without subsequent extraoperative language mapping were included, the rate of successful language lateralization was 17 (81%) of 21 overall. Table 2 shows the incidence of Wada testing broken down by age and language-mapping criteria. The difference in the success rate of language lateralization by Wada testing between children younger than 10.2 years and older children is statistically significant (χ2 test, p < 0.05).
Table 1. Patient data
|Patient no.||Age at onset||Seizure frequency||Age at surgery||Pathology||Handedness||Wada||Implant side||Language found|
| 1||2 yr||1–3/day|| 4.7||TS||Left|| ||Left||No|
| 2||5 yr||2/h|| 4.7||FCD||Right||Left||Left||Yes|
| 3||11 mo||3–4/wk|| 5.2||Tumor||Right|| ||Left||No|
| 4||5 yr||10/day|| 6.1||FCD||Right||Unable||Left||No|
| 5||3 yr||2–4/mo|| 6.2||Tumor||Left|| ||Left||No|
| 6||4 yr||4/mo|| 6.6||Encephalitis||Right|| ||Left||No|
| 7||4 mo||3–4/day|| 6.8||FCD||Right|| ||Left||No|
| 8||10 mo||2/mo|| 6.8||Stroke||Left|| ||Left||No|
| 9||15 mo||3–4/mo|| 7.4||MTS||Right||Unable||Right||No|
|10||3 yr||1-2/wk|| 8.1||MTS||Right||Left||Left||Yes|
|11||No discrete events|| 8.1||LKS||Right|| ||Left||Yes|
|12||3 yr||N/A|| 8.2||No data||Left||Unable||Left||No|
|13||8 mo||1/day|| 9.0||TS||No data|| ||Left||No|
|14||7 yr||2–3/day|| 9.3||FCD||Right|| ||Left||No|
|15||4 yr||12/mo|| 9.5||MTS||Right|| ||Left||No|
|17||6 yr||2–4/mo||10.2||TS||Right|| ||Left||No|
|19||5 yr||>1/day||10.6||No data||Right|| ||Left||Yes|
|20||9 yr||2–6/wk||12.1||FCD||Right|| ||Left||No|
|21||18 mo||20–30/wk||12.1||MTS||Right|| ||Left||Yes|
|22||6 yr||>1/wk||12.1||MTS||Right|| ||Left||Yes|
|23||11 yr||1–2/wk||13.1||MTS||Right|| ||Left||Yes|
|28||7 yr||2–3/day||14.0||MTS||Right|| ||Left||Yes|
|29||3 yr||3–4/wk||14.1||Tumor||Left|| ||Left||Yes|
|31||5 yr||2–3/wk||14.3||MTS||Right|| ||Left||Yes|
Although the value of a direct comparison between the success rates of Wada testing and stimulation mapping is limited because of the differing methods, testing paradigms, test outcomes (i.e., lateralization versus localization), and small sample sizes, the incidence of successful language lateralization with the Wada test (Table 2; four of seven subjects for a rate of 57%) was significantly greater (p < 0.05) than the incidence of successful language mapping in children younger than 10.2 years (Table 1; three of 16 patients for a rate of 19%). In our series, in one child (age 13.5), Wada testing failed to identify language, but localization was successful with cortical stimulation. Three children (ages 6.1, 7.4, and 8.2 at surgery) failed to have language positively identified through either Wada testing or cortical stimulation; however, in two of the children (ages 6.1 and 8.4), partial data from the Wada test suggested right hemisphere language, possibly explaining why cortical stimulation of the left hemisphere failed to identify language function.
Postsurgical language outcome was available in all but one case (patient 19). This was based on neurologic examination and on partial postoperative neuropsychological testing in eight patients. An expressive aphasia was noted at follow-up in patient 2 and a global encephalopathy in patient 25/27; however, these were attributed to progression of the underlying neurologic disorder in both cases. Both of these patients were evaluated with a neurologic examination alone.
- Top of page
- MATERIALS AND METHODS
This study found that the sensitivity of extraoperative language mapping is reduced in young children compared with adults, with a sharp transition to the adult range of sensitivity occurring at age 10. Wada testing, which focuses on lateralization of language rather than precise localization and uses a somewhat different testing paradigm from extraoperative neurostimulation at this institution, also shows significantly reduced sensitivity for the younger age group, but to a lesser degree. These findings demonstrate a previously undescribed stratification of pediatric functional-mapping patients by age group that clarifies the relationship of sensitivity to age, provides useful data for designing presurgical epilepsy evaluations, and may shed light on the timeline of language-network maturation.
Although both intraoperative and extraoperative language mappings by cortical stimulation have been performed in children undergoing brain resection for tumors or epilepsy (Berger et al., 1989; Hinz et al., 1994), evidence suggests that the sensitivity of the stimulation procedure in children is less than that for adults. Our findings are consistent with earlier studies of pediatric functional language mapping that show a relative paucity of findings in children (Jayakar et al., 1992; 1994; Ojemann et al., 2003). In addition, a lower incidence of motor identification by stimulation mapping, comparable to the results of language mapping, has been found (Jayakar et al., 1994; Chitoku et al., 2001). One study concluded that no difference exists in mapping results between children (aged 3–18) and adults (Wyllie et al., 1988). However, the subjects were mostly adolescents, with a mean age of 14.0; thus data from children younger than 10 years likely did not contribute significantly to the outcome.
Wada testing also appears to be less diagnostic in children than in adults. In addition to the mechanical limitations of this modality, including cross-hemispheric blood flow and uncertain distribution of amytal in the ipsilateral hemisphere (Hong et al., 2000), difficulties that are particular to children are experienced. Unilateral language has not been consistently identified in pediatric studies (Hinz et al., 1994; Jansen et al., 2002). In our case series, greater diagnostic yield was found for Wada test language lateralization in children older than 10, paralleling the threshold for neurostimulation mapping. However, obtundation interfering with testing was noted in four subjects (ages 5.9, 7.4, 8.3, and 13.4); indeed, for the children in this study, every Wada failure was ascribed to this phenomenon. The mean amytal dose given during injections that produces obtundation does not differ significantly from the overall numbers (Table 2), suggesting that it is not produced by dose–response kinetics. Concern over this effect, combined with a natural reluctance to order invasive procedures for children, likely accounts for the low rate of our preoperative Wada testing.
No postoperative language deficits directly attributable to the surgical resection were identified. Although it does not appear that a failure to localize language function had an adverse effect on outcome, it is possible that this led to more conservative resections than might otherwise have been carried out in some cases.
Several potentially confounding variables could explain negative language findings in the pediatric series reviewed here. First, language may be located in the contralateral hemisphere or may be bilaterally distributed. Although the majority of subjects in this series underwent implantation of the left, and presumed dominant, hemisphere, only 10 (33%) of the 30 underwent preoperative Wada testing for language lateralization. Thus contralateral language localization may have been present in some patients. Additionally, a higher incidence of left-handedness was noted in the younger age group (35%) compared with the older group (14%), suggesting increased atypical language representation in the younger group. For example, patient 9 did not undergo preoperative language lateralization, was right-handed, and had subsequent implantation and language mapping in the right hemisphere. Four patients had left hemisphere mapping who were left-handed, and one who was ambidextrous. Of the 16 children younger than 10.2 years, six (38%), compared with four (29%) of the 14 children 10.2 years or older underwent successful language lateralization preoperatively. In contrast, 14 (76%) of 18 adult controls underwent Wada testing as part of the presurgical epilepsy evaluation; the difference in the rate of Wada testing in adults compared with both groups of pediatric patients is statistically significant (χ2 test, p < 0.05). As a result, subjects in the adult control group had a higher pretest probability of identifying language. Although this may limit direct comparisons between the adult and pediatric populations in this series, the difference between rates of Wada testing in older and younger children was not statistically significant.
To assess whether our findings could have been affected by including children with lower pretest probabilities of positive language identification than the adult population, a subgroup comparison was performed for those subjects who were right-handed and had left-hemisphere implants (patients 2, 3, 6, 7, 10, 11, 14–26, 28, 30, and 31). In addition, we did not include the children who failed Wada testing but had a suggestion of right hemisphere language function, as well as the second procedure of patient 27. This calculation again showed a significant difference in the success rate of extraoperative language mapping between children aged 10.2 and younger and those older than 10.2 (p < 0.01; χ2 value, 8.96).
Several other factors, however, could confound the results. First, in some subjects, the electrode array may not have optimally sampled language areas. This factor, however, should not vary with the age of the subject and is just as likely to occur in adults. Second, the incidence of significant developmental delay in language could limit testing. Although evidence from fMRI studies of language localization in normal children suggests that language networks are well lateralized and localized by age 5 to 7 years (Ahmad et al., 2003; Gaillard et al., 2003; Sachs et al., 2003), the question arises whether cortical dysfunction or dysgenesis could directly affect the ability of neurostimulation to detect language sites. This hypothesis, however, was not borne out in our series, although a consistent, precise measure of language development by age was not available for the children in this study. For example, patient 11 was the second-youngest successful language mapping at age 8.1. This patient had Landau–Kleffner syndrome with severe language impairment, whereas patient 7 at 6.8 years of age had a negative mapping despite good cooperation, normal language function, right-handedness, wide grid coverage of the left lateral and inferior frontal lobe extending posteriorly to include the primary motor area, and repeated testing at maximum current and with varying pulse width and frequency settings.
Last, the incidence of disorders of cortical migration, which is more likely to affect language representation in patients younger than 10.2 (47%), was greater than the incidence in the older age group. Because the rate of successful language mappings among children without migrational disorders was two of nine (22%) in the younger group and 11 of 11 (100%) in the older group, the mismatch in etiologies cannot fully account for the difference in success rates between the two groups. However, it is possible that the cumulative increase in risk factors for atypical language representation seen in the younger age group (increased left-handedness, higher percentage of migrational disorders, younger age at onset) partially accounts for the lower rate of success with cortical stimulation.
Even taking these factors into account, we believe that this study reveals reduced sensitivity of language mapping by cortical stimulation in children younger than 10 years. A corresponding reduction in sensitivity of the Wada test occurs in this age group, although it is a comparatively smaller effect. Given the low sensitivity of either of the two diagnostic modalities studied, it is reasonable to conclude that preoperative language lateralization, particularly with noninvasive techniques such as functional MRI and/or magnetic source imaging (MSI), should be emphasized in the presurgical evaluation of children younger than 10 years. BOLD fMRI studies of both verbal fluency tasks (Gaillard et al., 2003) and auditory comprehension (Ahmad et al., 2003) indicate that localization and lateralization of language in both the frontal and temporal lobes can be obtained with noninvasive testing in children as young as 5 years of age.
The failure of electrical cortical stimulation to elicit functional findings consistently in young children has been attributed to continuing neurodevelopment, in particular, reduced myelination and a greater proportion of small fibers. Bipolar electrical stimulation produces a charge density under the selected electrodes and including the tissue between them, which is proportional to the applied current times the pulse duration, and inversely proportional to the area of the electrodes (Oostendorp and van Oosterom, 1991; Jayakar, 1993). Cortical response is produced by a charge density sufficient to trigger action potentials in the affected tissue. This minimal density lies along a curve of pulse current versus pulse duration (Jayakar, 1993). The effect of reduced myelination is to shift the curve to the right, thus increasing the charge density required for a clinical response. Earlier studies have confirmed that the stimulation thresholds required to elicit motor findings decrease with age (Jayakar et al., 1992; 1994; Chitoku et al., 2001). Thus the required stimulation threshold may be increased in children beyond the afterdischarge threshold, the capability of the stimulator used for the procedure, or the known safety limits. The sharp increase in successful language mapping at approximately age 10 suggests that neurophysiologic development reaches a threshold at this age that permits effective stimulation with the clinical stimulation parameters in use at this institution.
Strategies to increase the charge density applied have improved the sensitivity of the mapping procedure in children. Cases in which currents upto 20.5 mA have been required to identify functional cortex have been reported (Chitoku et al., 2001; Ojemann et al., 2003). Another method of increasing the charge density applied during mapping would be to use smaller subdural grids with reduced electrode diameters and interelectrode distances.
However, with increasing current density, the potential for tissue injury increases, and safety limits established for adults (Gordon et al., 1990) do not take into account the increased parameters that may be used in children. A proposed algorithm to increase stimulation parameters in a stepwise fashion to converge them to the chronaxie, or the minimum point along the pulse current–duration curve, results in maximizing the cortical response while minimizing the charge density applied (Jayakar et al., 1992). Improved motor mapping results by using this paradigm have been reported (Duchowny and Jayakar, 1993). However, this method requires a clinical stimulator with the capability of generating pulse durations of up to 2 ms.
Our study demonstrates the limits of the sensitivity of direct cortical stimulation for language localization in children younger than 10.2 years. Although this series has a limited number of patients with both preoperative Wada testing and subsequent cortical stimulation, the data demonstrate decreased sensitivity of both direct cortical stimulation using subdural grid electrodes and the Wada test in identifying language in younger children, with the Wada test being somewhat more sensitive in this age group. To maximize the pretest probability of language identification with cortical stimulation, we recommend preoperative language lateralization using noninvasive tests such as fMRI and magnetoencephalography (Papanicolaou et al., 2004). It may also be possible to increase the ability of Wada testing to lateralize language by modifying the pharmacologic protocol. In addition, strategies for modifying stimulation parameters may yield improved results; however, further investigation into the effects of increasing charge density must be done to ensure safety.