The Effects on Cognitive Performance of Tailored Resection in Surgery for Nonlesional Mesiotemporal Lobe Epilepsy


Address correspondence and reprint requests to Dr. F.S.S. Leijten at University Medical Center Utrecht, Department of Clinical Neurophysiology F02.230, Heidelberglaan 100, 3584 CX Utrecht, The Netherlands. E-mail:


Summary: Purpose: Mesiotemporal lobe epilepsy (MTLE) can be treated with different surgical approaches. In tailored resections, neocortex is removed beyond “standard” margins when spikes are present in the electrocorticogram. We hypothesized that these larger resections are justified because spiking neocortex is dysfunctional. This would imply that in patients with spikes (a) postoperative cognitive performance is not affected, and (b) preoperative performance is worse than without spikes.

Methods: We studied 80 operated-on MTLE patients with pathologically confirmed nonlesional hippocampal sclerosis. All patients were left-sided language dominant and underwent cognitive tests 6 months pre- and postoperatively. A repeated measures analysis of variance (ANOVA) was performed, looking for within- and between-subjects interactions with presence of intraoperative neocortical spikes.

Results: Intraoperatively, neocortical spikes were present in 61% of patients. Improved postoperative cognitive outcome was seen only in left-sided patients with spikes. Their performance IQ (PIQ) increased by 8.1 points (95% confidence interval, 3.8–12.3; p = 0.02), and visual naming latency by 12.8 s (95% CI, 2.1–23.5; p = 0.07). Conversely, in left-sided patients without spikes, naming latency declined by 7.5 s (95% CI, −2.3–17.2; p = 0.07). Preoperative scores were comparable except for a 15.3-point (95% CI, 0.1–30.5; p = 0.02) lower VIQ in left-sided patients without spikes.

Conclusions: Tailoring does not harm cognitive performance and is, in left-sided MTLE, associated with postoperative improvement. Left-sided MTLE without neocortical spikes has lower verbal scores, which tend to decline after standard resection and may represent a special pathophysiologic entity.

The syndrome of nonlesional mesiotemporal lobe epilepsy (MTLE) is well recognized and a principal target for epilepsy surgery worldwide (1). The syndrome is defined by a set of clinical, interictal, and ictal EEG and magnetic resonance imaging (MRI) criteria. Resection of the amygdala and the hippocampus is the mainstay of its surgical treatment, as archicortical structural changes in the mesial temporal lobe (sclerosis and atrophy) are held critical to seizure propagation in the majority of cases (2). However, coexisting independent spiking in the basal and lateral temporal neocortex have been known in the EEG literature for a long time (3). It has been proposed that in some of these cases, mesiotemporal pathology may be the consequence rather than the cause of epilepsy. Some authors confirmed the presence of microscopic neocortical abnormalities that were not seen in MRI (4). The electroclinical characteristics of these neocortical temporal lobe epilepsies with secondary mesiotemporal involvement may be indistinguishable from MTLE proper (5).

This has led to three different surgical approaches toward MTLE: (a) Selective amygdalohippocampectomy without neocortical resection is the minimal form of resective surgery, requiring invasive monitoring to make sure that neocortical foci do not play a role in seizure generation; (b) to avoid invasive monitoring, other centers include in the resection a “standard” amount of neocortex, as is deemed safe (6)—that is, insofar as this tissue is unlikely to support language function; (c) the eldest approach, however, is to “tailor” the neocortical resection beyond the standard margins in the individual patient.

This intraoperative tailoring of the neocortical resection is led by acute, intraoperative electrocorticography (ECoG) (7). Because of conflicting results in outcome studies in MTLE (8–11), this practice has received considerable criticism by proponents of the “no more than standard” resections (12). Standardized procedures typically involve removal of 3–4 cm of anterior temporal neocortex, whereas tailoring could lead to removal of ≤7 cm, depending on the presence of neocortical spikes and eloquent functions. Its equivocal benefit and this trend toward larger resections has been a concern for the tailored approach toward MTLE (13,14). Larger resections may produce cognitive deficits, especially in the language-dominant temporal lobe (15–17). This will be true even with prior or intraoperative intracranial language mapping (6).

Conversely, the presence of spikes in the EEG, or the ECoG for that matter, may be a sign of network dysfunction (18). Such an interictal epileptic condition can be associated clinically with decreased performance (19) and so-called transient cognitive impairments (20). Cognitive improvement may follow the treatment of these interictal abnormalities, even when seizures are scarce. The idea is that a local network dysfunction may interfere not only with local function but also with the global network, and, vice versa, that removal of a diseased part of the network may allow the expression of compensation mechanisms. Network changes and dysfunction may also apply to the adult brain, especially if the etiology of MTLE is still in doubt and may go back to the neonatal period. Indeed, improvement on IQ scales has been described after temporal lobe resection in adults (17). In this view, gross removal of temporal neocortex may not be harmful per se (21) and may even be the best treatment option in selected patients.

We decided to test the hypothesis that a tailored resection in MTLE is not detrimental (maybe even beneficial) to performance, because its rationale pertains to the removal of dysfunctional brain. We therefore investigated retrospectively the pre- and postoperative cognitive test performance in a cohort of consecutive MTLE patients who underwent a tailored resection in the National Dutch Collaborative Epilepsy Surgery Program. According to the hypothesis, we would expect to find (a) better postoperative neuropsychological test scores in resections tailored by the presence of spikes than in standard resections without spikes, and (b) lower preoperative test scores in patients with extensive neocortical spiking in their ECoGs.

Because most tests involve language processing, the effects would be expected to be most pronounced in the language-dominant temporal lobe.



We were able to study 80 of 86 consecutive patients with (a) a pathologically confirmed, clinical, electrophysiologic and MRI diagnosis of unilateral, nonlesional MTLE with mesiotemporal sclerosis, (b) with unilateral, left-sided language dominance as determined by the Wada test. Bilateral or atypical (right-sided) language dominance was excluded because these cases have different cognitive profiles for reasons that are not fully understood (22,23). The 80 patients in our study were consecutive cases in an 11-year period between March 1991 and March 2002, from the database of the Dutch Collaborative Epilepsy Task Force that coordinates all epilepsy surgery in the Netherlands (24). One patient died within half a year postoperatively; in two patients, intraoperative ECoG failed technically, and in three patients, the neuropsychological data were incomplete.

Preoperative electrophysiologic investigations included a prolonged video-EEG seizure recording on which a consensus conclusion was reached by a team of three experienced electrophysiologists. A dedicated MRI protocol was used as described elsewhere (25), including coronal fluid-attenuated inversion recovery (FLAIR) images of the temporal lobes. The Wada intracarotid amytal procedure was performed bilaterally in all patients. According to this test, patients were categorized as having left or right hemisphere language function, or both. Only patients with left-sided language dominance were included. Surgery included full amygdalohippocampectomy in all patients, resection of the anterior temporal lobe of 3–4 cm, and further resection only by tailoring with ECoG during operation. During operation, the final extent of the resection was measured in centimeters for each temporal gyrus, and a mean resection width was calculated. After resection, the en bloc hippocampus specimen and anterior temporal white matter and neocortex were scrutinized by our neuropathologist. If a specimen contained focal cortical dysplasia or tumor, the patient was excluded from the analysis. A specialized team of two neurosurgeons (including P.V.R.), three neuropsychologists (including W.A. and J.V.), one neuropathologist, two neuroradiologists, and four neurophysiologists (including F.L. and A.v.H.) were involved in the care of all patients.

Psychological tests

Patients underwent a standard neuropsychological test battery 6 months before and 6 months after surgery. For the purposes of this study, eight results from five validated different tests were examined:

  • – The Dutch version of the Wechsler Adult Intelligence Scale (WAIS) with Verbal IQ (VIQ) and Performance IQ (PIQ). The used version is the 1970 WAIS. Probably because of the Flynn effect (26), our norm population has a mean IQ of half a standard deviation above the expected mean of 100.
  • – The 15 Words Test, a verbal learning test. This is a Dutch adaptation of Rey's Auditory Verbal Learning test (27). It consists of five successive presentations of a list of 15 words followed by free recall on each trial. After interference tasks have been performed for half an hour, retention of the list is examined. The test yields two measures for the purpose of the study: the total words recalled across the five trials as a measure of list learning or acquisition (immediate recall), and the long-delay free-recall score as a measure for consolidation and retrieval (delayed recall). To minimize retest effects, an equivalent alternate word list was used at follow-up.
  • – The Visual Naming Task, a 15-item test of black-and-white drawings. The total number of naming errors of 15 pictures is scored, as well as the total time in seconds to name all these 15 items.
  • – The Digit Span Forward, as a test of auditory alertness and short-term audioverbal recall.
  • – The Rey-Osterreith Complex Figure Task, in which the subject first has to copy a figure and after 10 min is asked to reproduce the figure just copied. The delayed-recall score was used in our study. This test is supposed to measure long-term visual memory (28).


During surgery, a regular three-stage ECoG (each of 15 min) was performed after discontinuation of propofol anesthesia. Before resection, 12–16 saline wick electrodes or a 4 × 5 stainless-steel electrode grid (Ad-Tech, Racine, WI, U.S.A.) were placed over the lateral surface of the inferior, middle and superior temporal gyrus. A 1 × 7 strip electrode (Brain-Electronics, Houten, The Netherlands) was introduced subdurally underneath the temporal lobe toward the hippocampus, at 4 cm from the temporal pole (stage 1). Thereafter, a standard 3- to 4-cm anterior temporal resection was performed, and the temporal horn of the lateral ventricle was opened, and another 1 × 7 strip electrode was introduced inside, with the other strip electrode underneath the temporal lobe (stage 2). Thus direct, longitudinal recording becomes possible over the hippocampus proper (29). After this, en bloc amygdalohippocampectomy was performed. Then final tailoring was performed by ECoG (stage 3) by using the same lateral and subtemporal recording technique as in stage 1. When resections were on the dominant side, intraoperative language mapping in the awake patient was usually done by using electrocortical stimulation during a computerized picture-naming task. If patients were unable to undergo this procedure, the neocortical resection was limited to 3 cm. Data were digitally stored (BrainStar; Schwind Co., Germany) at a sample rate of 512 Hz and filter settings of 0.1–140 Hz. The recordings were later reviewed off-line in a blinded manner by one of the authors (F.L.). Spikes were considered neocortical when recorded exclusively over the lateral temporal surface at stage 1 (before anterior resection). They were considered as hippocampal when seen at the tip of the subtemporal strip in stage 1 and confirmed by intraventricular strip recordings in stage 2. Anterior neotemporal spikes were classified as hippocampal if they were propagated from the mesial structures (30), and as neocortical, when they were independent.

Group definition

We defined two groups on the basis of the ECoG findings (viz., one group of patients in whom the neocortical resection was supported by the presence of neocortical spikes in the ECoG, and another group in whom the neocortical resection was part of a standard approach, but not supported by the presence of spikes). Each group was subdivided into left- and right-sided cases.


Between-group comparisons were carried out with two-tailed independent sample t testing (equal variances assumed after F testing), or with the nonparametric Mann–Whitney U test in case of categoric variables. Interactions between side of operation, presence of ECoG spikes, and test changes before and after operation between and within subjects was analyzed by a repeated-measures approach (ANOVA) for each test. The distribution of the each data set was checked for normality with box plots including median, extreme values, and outliers. In one case of clearly nonnormal distribution (i.e., the Visual Naming Errors test), data were reanalyzed nonparametrically. This was achieved by calculating the difference between postoperative and preoperative scores, categorizing these in three groups (<0, 0, and >0) and performing Pearson's χ2 test on each 3 × 2 frequency table for side and presence of spikes. Significance was assumed if the p value was <0.05. Significant within-subjects effects were shown as means with standard deviation. Significant between-subjects effects were expressed by 95% confidence intervals of the means. All calculations were made by using SPSS software version 11.5 (SPSS Inc., Chicago, IL, U.S.A.).


Overall characteristics

Sufficient data were available for 80 patients with pathologically proven, nonlesional mesiotemporal sclerosis and left-sided language dominance. The intraoperative ECoG showed neocortical spikes in 49 (61%) patients. In Table 1, the characteristics of these 80 patients are shown, according to side of operation and the presence of neocortical spikes on intraoperative ECoG. Because all patients were left-side language dominant, “right side” also may be read as “non–language dominant.” No statistical differences were found between the four groups with regard to sex, age at first nonfebrile seizure, or age at operation. More patients with right-sided surgery, however, had a childhood seizure onset (p = 0.06). Mean duration of epilepsy at operation was 23 ± 9 years for left-sided, and 27 ± 12 years for right-sided patients (p = 0.14). As expected, extent of neocortical resection differed between sides: on average, 3.6 ± 0.9 cm on the left and 4.3 ± 1.2 cm on the right (p = 0.002). Extent of resection also differed according to the corticographic findings: 3.6 ± 1.2 cm without neocortical spikes and 4.3 ± 1.0 cm in the presence of spikes (p = 0.006). After 1 year, outcome with respect to seizures was excellent, with 59% of patients being completely seizure free (Engel class IA) (31), and an additional 20% being essentially seizure free except for auras, or only having had seizures immediately postoperatively or after discontinuation of medication (Engel IB–D). Thus 79% of patients were in the best outcome category (Engel I). Another 14 patients had rare seizures (Engel II). Only three (4%) patients were considered surgical failures (Engel III), still with a ≥75% reduction in seizures. All three patients had neocortical spikes on their ECoGs.

Table 1. Patient characteristics according to side of resection and presence of lateral neocortical spikes on intra-operative electrocorticogram (means ± standard deviation)
Side of operation
(n = 80)
No spikesSpikesNo spikesSpikes
N (male/female)16 (7/9)18 (8/10)15 (6/9)31 (17/14)
Age (years) at seizure onset11 ± 711 ± 1011 ± 128 ± 7
N seizure onset ≤ 6 years5 (31%)7 (39%)9 (60%)17 (55%)
Age at operation (years)36 ± 933 ± 936 ± 1036 ± 11
Extent of resection (cm)3.4 ± 1.13.7 ± 0.83.8 ± 1.44.6 ± 1.0
Postoperative outcome (31) after 1 year
 N completely seizure free (IA)810920
 N auras only or postoperative seizures only (IB,C,D)5335
 N rare seizures (II)3434
 N worthwhile improvement (III)0102

Within-subjects effects

No clear general changes in VIQ were seen after operation (p = 0.14; Fig. 1). No interaction for VIQ occurred with side of operation (p = 0.27), presence of ECoG spikes (p = 0.52), or both (p = 0.64); see Table 2. General PIQ, however, improved statistically after operation (p = 0.008; Fig. 1), although the overall effect was small [2.5 points; 95% confidence interval (CI), 0.4–4.5, from an overall 107.7 ± 15.4 to 110.1 ± 17.0]. Here, the interaction analysis shows that the increase was due to an 8.1-point (95% CI, 3.8–12.3) improvement that was confined to the group of left-sided patients with neocortical spikes (Fig. 1 and Table 2). To test whether this change in PIQ was due to an improvement of attention, we analyzed the scores on the WAIS Substitution subtest, which did not show any change.

Figure 1.

Estimated marginal means of cognitive test results before (“pre”) and after (“post”) surgery, according to side of operation and presence of neocortical spikes on ECoG.

Table 2. Within-subjects effects (means and standard deviation) of presence of intra-operative neocortical spikes on cognitive test parameters
N = 80LeftRight
No spikesSpikesNo spikesSpikes
Verbal IQ
Preoperative92.3 ± 17.0107.6 ± 18.7 105.2 ± 17.3 107.5 ± 13.6 
Postoperative91.9 ± 15.3108.5 ± 18.0 106.7 ± 17.7 109.2 ± 14.4 
Performance IQ
Preoperative106.3 ± 16.6 106.7 ± 16.2 107.0 ± 14.4 109.3 ± 15.5 
Postoperative107.3 ± 17.7 114.8 ± 18.6 109.9 ± 18.5 109.0 ± 15.0 
Immediate 15-Word Recall
Preoperative43.5 ± 13.246.3 ± 11.347.0 ± 7.7 45.0 ± 9.3 
Postoperative35.8 ± 11.944.1 ± 10.947.9 ± 7.4 46.3 ± 6.7 
Delayed 15-Word Recall
Preoperative8.6 ± 4.08.1 ± 4.09.0 ± 4.38.9 ± 3.3
Postoperative6.7 ±3.57.1 ± 4.110.4 ± 3.5 9.8 ± 2.7
Visual Naming Errors
Preoperative0.3 ± 0.70.3 ± 0.70.3 ± 0.50.3 ± 0.7
Postoperative1.3 ± 1.30.7 ± 1.20.3 ± 0.70.0 ± 0.0
Visual Naming Latency
Preoperative18.8 ± 10.324.4 ± 31.814.3 ± 5.1 13.1 ± 2.9
Postoperative31.6 ± 29.017.0 ± 5.4 16.7 ± 5.4 13.8 ± 4.8
Digit Span Forwards
Preoperative4.9 ± 1.65.8 ± 1.45.7 ± 1.65.6 ± 1.1
Postoperative5.3 ± 1.45.3 ± 1.35.3 ± 1.05.8 ± 1.4
Rey's Complex Figure
Preoperative17.2 ± 4.9 18.8 ± 7.6 15.4 ± 5.2 14.8 ± 5.2 
Postoperative20.5 ± 6.1 20.1 ± 7.2 18.9 ± 6.9 18.0 ± 4.7 

The six items of the more specific tests also were analyzed (Table 2). Most of these tests did not show an overall change after operation, except for an improvement in the delayed reproduction of Rey's Complex Figure. Interactions with side of operation were found for Immediate and Delayed 15-Word Recall, and Visual Naming Errors (Table 3). Performance on both word-recall tests declined in left-sided resections and remained about the same in right-sided ones. Visual Naming Errors increased postoperatively in left-sided cases, irrespective of the presence of spikes, and tended to decrease in right-sided cases, especially when neocortical spikes were present (but this was not statistically significant).

Table 3. Significant interactions (and trends), expressed in p-values, between consecutive neuropsychological test results (dependent variables) and side of operation and presence of neocortical spikes (independent variables); *analyzed nonparametrically as explained in the Methods
InteractionWithin-subjects effectsBetween-subjects effects
Moment (pre/post

Side × moment

Spikes × moment
Side × spikes
× moment



Side × spikes
VIQ (0.06)0.02(0.07)
PIQ0.008 0.02 
Immediate Recall(0.06)0.003 (0.08) 
Delayed Recall 0.002 0.03 
Visual Naming Latency 0.03(0.07)0.02 
Visual Naming Errors  0.005* 0.02 
Rey's Figure<0.001 (0.09) 
Digit Span Forwards 0.03 

Interaction with the presence of spikes was seen in the Visual Naming Latency and Digit Span Forward (Fig. 1 and Table 3). In the presence of spikes, both measures slightly improved, whereas they deteriorated in the absence of spikes. The latency improved from a mean 17 to 15 s in the presence of spikes and declined from 16 to 23 s in the absence of spikes. This effect tended to be strongest in left-sided cases (Fig. 1). On the left, the improvement was a mean 12.8 s (95% CI, 2.1–23.5 s) in patients with spikes. A mean 7.5-s decline (95% CI, −2.3–17.2 s) was noted in left-sided cases without spikes. However, standard deviations were large. The results of Digit Span Forward could not be captured in simple categories (Fig. 1); although a significant interaction was found, the effects were very small (0.2–0.5 in either direction around a total mean of 5.5 digits recalled before and after surgery).

Between-subjects effects

Clear differences were seen in VIQ between patients with and without spikes. With neocortical spikes, left- and right-sided patients together showed a mean VIQ of 107.5 (95% CIs: 102.8–112.3) preoperatively and 108.8 (104.1–113.6) postoperatively. Without spikes, preoperative VIQ was 98.8 (93.0–104.6) and postoperative VIQ, 99.3 (93.5–105.0). The mean difference between the groups with and without spikes, therefore, is 8.7 preoperatively and 9.5 IQ points postoperatively. Patients with left-sided MTLE tended to have a lower preoperative VIQ by a mean 6.3 points compared with right-sided MTLE, and this difference increased to 7.7 points postoperatively, which reached almost statistical significance (p = 0.06). In the interaction analyses, it seems that the subnormal pre- and postoperative VIQ is confined to left-sided patients without spikes (Fig. 1). On the left, patients without spikes had a 15.3-point (95% CI, 0.1–30.5) lower VIQ than those with spikes in the preoperative situation, and a 16.6-point (95% CI, 1.4–31.8) postoperative difference. Patients with spikes on the left seem to perform like their right-sided counterparts (Fig. 1).

No between-subjects differences were found in preoperative PIQ with regard to side of operation (p = 0.99), presence of spikes (p = 0.52), or both (p = 0.65). Within the other tests, no further interactions were noted with presence of spikes, only with side of operation. Parameters for word recall and visual naming showed lower scores in left-sided patients (Table 2).

Other analyses

Hippocampal spikes on ECoG were present in 62 patients and absent in nine. In the other nine patients, no definite conclusion was possible on the presence of hippocampal spikes, mainly for technical reasons. A repeated measures ANOVA analysis with hippocampal spikes in the model did not yield any significant interaction.

The repeated measures ANOVA also was performed on all tests, with operation size as a covariate. No interaction was found with operation size in any of the test results.


About 60% of patients with nonlesional MTLE syndrome show independent neocortical spikes in the intraoperative ECoG. According to our hypothesis, we predicted that removal of this dysfunctional (i.e., spiking neocortex) would not negatively affect cognitive outcome and might even improve it. The latter is supported by the marked improvement in PIQ in left-sided cases with spikes (Table 2). With regard to VIQ, changes are not clearly affected by side of operation or presence of spikes—not even by extent of resection. After operation, visual-naming latencies improved slightly when neocortical spikes were present and deteriorated when no spikes were present. Thus evidence exists in line with the idea that removal of temporal tissue that produces spikes is beneficial, or, reversibly, that it may be harmful to remove neocortex that does not spike.

In the second part of our hypothesis, we stated that patients with spikes would be expected to have lower preoperative test scores, especially when the left, language dominant temporal lobe was involved. Indeed, scores on VIQ, word-recall tests, and visual naming parameters were lower in left-sided cases. However, the only significant interaction with presence of neocortical spikes in the ECoG was that without spikes, VIQ was lower. Patients with left-sided MTLE without spikes in the ECoG, had the lowest VIQ before and after operation (Table 2), well below the normal range (normal VIQ in the general population was 108 with this version of the WAIS). Thus this part of the hypothesis seems refuted by the results.

Reconciling the paradoxical findings is not straightforward. First, in left-sided patients with spikes, it is the PIQ and not the VIQ that improves after resection. Second, the already subnormal VIQ in patients without spikes on the left does not improve postoperatively. Maybe the presence of neocortical spikes is, apart from its epileptic significance, a marker of plasticity on a wider scale. Vice versa, the absence of spikes may be the expression of a more localized disease with more predominant inhibition that is somehow not relieved by standard resection. We know from positron emission tomography (PET) studies that MTLE may have different intra- and extratemporal neocortical metabolic effects (32), and from functional MRI, that network accessibility outside the temporal lobe may be affected (26). Combining ECoG and cognitive test results with PET and functional MRI may in the future shed more light on the different IQ profiles, neocortical spiking, and effects of operation in MTLE.

Comparing our findings with other cognitive-outcome studies of temporal lobe epilepsy is difficult. We are not aware of other such studies that take the intraoperative ECoG into account. Moreover, most studies mix MTLE and lesional temporal lobe epilepsy cases (33,34). We have used very strict criteria, including pathologic examination of surgical tissue, to ensure that our population had “pure” MTLE. Some studies compare different operation techniques (16,33,35), but although these rely on some sort of invasive seizure monitoring, they do not offer data on ECoG findings. Almost all studies find that memory and language test performance is affected by left- but not right-sided epilepsy and surgery (36–40). The finding of a postoperative improvement in performance (but not verbal) IQ in left-sided temporal lobe epilepsy is in line with recent studies in children (41,42). Our study shows that the effect is restricted to left-sided patients with neocortical spikes in their ECoGs. Whether this improvement is clinically meaningful remains to be seen; we did not include a control group that would have allowed us to calculate reliable change indices (43).

The overall picture that arises from our study is that two different functional disease states exist in left-sided MTLE depending on the intraoperative presence of neocortical spikes, irrespective of clinical, MRI, or pathologic findings. One group of patients (about half of left-sided MTLE) do not show neocortical spikes on ECoG. This group has a worse VIQ from the outset (and below the average of the general population), does not improve after standard anterior resection in any of the tests we used, and even shows a 30% decline in visual naming latency. The other group does show neocortical spikes, has a normal VIQ that is unaffected by operation, and is rewarded by a significant improvement in PIQ after operation. This is even in spite of the fact that more cortex was removed in these patients with spikes.

Of course, this study is hampered by the fact that it was not designed as a test of tailoring versus nontailoring. It is therefore impossible to prove that tailoring of the neocortical resection was a superior treatment in left-sided MTLE. It is clear that tailoring is not harmful (and maybe even beneficial). It is thus untrue that large neocortical resections per se pose cognitive threats. A suggestion has been voiced that removing nonspiking neocortex is harmful with regard to visual naming latency. This can be taken as a support for tailoring of resections in left-sided MTLE. Conversely, we expected that the initial VIQ would be worse in patients with spikes, whereas the reverse is true. When all evidence is taken together, questions remain whether we should regard spiking neocortex as “dysfunctional” in all respects. Our findings may prompt further, prospective studies to see if the tailoring approach is really superior to the “standard resection” without tailoring. This would be especially interesting if tailoring would lead not only to larger neocortical resections, but also were allowed to restrict resections to fewer centimeters than used in a standard procedure, in case no intraoperative neocortical spiking were found.

Neocortical spiking is important in left-sided MTLE patients, as we have shown with ECoG. This raises the question whether this is not already evident from the (preoperative, noninvasive) interictal EEG. We think that the contribution of interictal surface EEG is essentially different from that of the ECoG. In spite of the disadvantages of intraoperative ECoG (limited recording time, anesthetic effects), ECoG has some strengths that the surface interictal EEG does not have. In ECoG, it is possible to show dependency of (anterior) neocortical spikes on a hippocampus generator [the “leading spike” (44)] and discern these from independent neocortical spikes. Furthermore, ECoG spikes may be more local than can be picked up by a skull electrode, which requires a minimal area of synchronous spiking for detection (45). Very few studies, conversely, have formally addressed the relation between intraoperative ECoG and standard EEG. One study concluded that “significant agreement between the broad topography of dominant interictal foci” was found and that “the number of epileptiform discharges in intraoperative preresection ECoGs substantially exceeded that of preoperative scalp EEGs” (46). In our experience, interictal spikes in pure MTLE syndrome are indeed difficult to obtain with preoperative EEG and even magnetoencephalography (MEG) and cannot be localized in a meaningful way (47). When intraoperative ECoG was compared with prolonged ECoG with subdural strips, interictal spike foci correlated well (46). The dominance of one focus could be different in both recordings; however, this is not relevant to our study, as we did not look for focus dominance.

We wonder whether the effects we see from the presence or absence of intraoperative neocortical spikes are restricted to left-sided cases. The bias here is that most tests we used, even the PIQ, rely heavily on verbal instructions. The only test that probably depends predominantly on right hemisphere function is the Rey Figure reproduction. Further studies might include more tests of visuospatial skills and orientation.


Left-sided MTLE is not one disease entity, in spite of similar clinical, ictal EEG, MRI, and neuropathologic findings. Intraoperative ECoG findings have consequences for cognitive performance on several psychological tests in patients who undergo a tailored resection because of left-sided, nonlesional MTLE. Absence of spikes is associated with a subnormal preoperative VIQ. This finding cannot be understood by the notion that only spiking neocortex is dysfunctional. The absence of independent neocortical spikes is associated with a postoperative decline of visual naming latency in left-sided cases. A marked increase in PIQ is seen only in left-sided patients with neocortical spikes, who were maximally tailored. These findings support, but do not prove, the usefulness of tailoring.


Acknowledgment:  We thank Ingeborg van der Tweel from the Department of Biostatistics, University of Utrecht, for her help.