Good surgical outcome in discordant ictal EEG-MRI unilateral mesial temporal sclerosis patients


Address correspondence to Luiz H. Castro, Al. Lorena, 983 apt. 82, Sao Paulo, Brazil. E-mail:


Purpose: Video electroencephalography (vEEG) monitoring of patients with unilateral mesial temporal sclerosis (uMTS) may show concordant or discordant seizure onset in relation to magnetic resonance imaging (MRI) evidence of MTS. Contralateral seizure usually leads to an indication of invasive monitoring. Contralateral seizure onset on invasive monitoring may contraindicate surgery. We evaluated long-term outcome after anteromesial temporal lobectomy (AMTL) in a consecutive series of uMTS patients with concordant and discordant vEEG findings, uniformly submitted to AMTL on the MRI evidence of MTS side without invasive monitoring.

Methods: We compared surgical outcome of all uMTS patients undergoing vEEG monitoring between January 1999 and April 2005 in our service. Discordant cases were defined by at least one seizure onset contralateral to the MRI evidence of MTS. Good surgical outcome was considered as Engel's class I. We also evaluated ictal SPECT concordance to ictal EEG and surgical outcome.

Results: Fifty-four patients had concordant (C) and 22 had discordant (D) scalp EEG and MRI. Surgical outcome was similar in both groups (C = 74% versus D = 86%). Duration of follow-up was comparable in both groups: C = 56.1 ± 20.7 months versus D = 59.8 ± 21.2 months (p = 0.83, nonsignificant).Discordant single-photon emission computed tomography (SPECT) results did not influence surgical outcome.

Discussion: Surgical outcome was not influenced by contralateral vEEG seizure onset or contralateral increased flow on ictal SPECT. Although vEEG monitoring should still be performed in these patients, to rule out psychogenic seizures and extratemporal seizure onset, a potentially risky procedure such as invasive monitoring may not only not be indicated in this patient population, but may also lead to patients erroneously being denied surgery.

Unilateral mesial temporal sclerosis (uMTS) is a surgically remediable syndrome (McIntosh et al., 2001). When presurgical video electroencephalography (vEEG) yields concordant data—electrographic seizure onset on the same side of the lesion—patients are referred for surgical treatment without further testing. Problems arise when noninvasive vEEG monitoring data shows EEG seizure-onset contralaterally to the magnetic resonance imaging (MRI) evidence of MTS (Williamson et al., 1993; King et al., 1997; Diehl & Lüders, 2000).

Discordant vEEG-MRI data may be associated with worse postoperative seizure control (Jack et al., 1992; Schulz et al., 2000), but this is not a consistent finding (Hirsch et al., 1991; Sirven et al., 1997). Discordant MRI-vEEG data on presurgical evaluation in uMTS can indicate either a contralateral structural epileptogenic abnormality, not detected by MRI (Babb, 1991) or false vEEG lateralization, resulting from rapid contralateral ictal activity spread, not necessarily determining a worse surgical outcome (Mintzer et al., 2004). Different interpretations of this phenomenon may influence surgical decision in these cases. Ictal SPECT studies could contribute to address ictal vEEG false lateralization.

uMTS patients with discordant MRI-noninvasive vEEG data usually undergo invasive monitoring to determine seizure onset side (Diehl & Lüders, 2000; Spencer, 2002). If contralateral seizure onset is seen in invasive studies, surgery may be contraindicated (Lieb et al., 1981; So et al., 1989; Hirsch et al., 1991; Sirven et al., 1997; Diehl & Lüders, 2000). Resective surgery on the MTS side, regardless of discordant MRI-vEEG data is also a justifiable option, since lesion resection is the most important prognostic factor for a seizure-free outcome (Berkovic et al., 1995).

We compared long-term surgical outcome in a consecutive series of uMTS patients with concordant and discordant MRI-noninvasive vEEG monitoring, uniformly submitted to temporal lobe resection on the lesion side, without invasive studies. We also analyzed the contribution of ictal SPECT studies in predicting surgical outcome in vEEG-MRI discordant cases.


Patient selection

We identified all patients with uMTS who underwent presurgical evaluation in the Epilepsy Surgery Program at the Department of Neurology at Hospital das Clínicas—Universidade de São Paulo—Brazil in the period between January 1999 and April 2005.

Patients were included if they had MRI-documented uMTS and no other MRI lesions, except mild white matter signal changes or up to five small calcified lesions, located outside the temporal lobe (suggestive of inactive neurocysticercosis) and if they had undergone noninvasive vEEG monitoring with at least two recorded seizures and had undergone surgery during that period.

Patients were excluded if ictal or interictal extratemporal activity was seen during vEEG, if they had vEEG documented nonepileptic psychogenic seizures, if patients had a full-scale IQ less than 70, or if patients were operated on by a surgeon other than the neurosurgeon of the epilepsy group in our service.

Surgical treatment decision

After noninvasive vEEG monitoring, all patients' data were analyzed in a seizure conference, where neurophysiological (ictal and interictal vEEG), MRI, and SPECT studies were analyzed by epileptologists, a neuroradiologist, a nuclear medicine specialist, and a neurosurgeon. MTS was defined as unilateral amygdala-hippocampus-parahippocampal complex atrophy on T1-weighted images and/or increased signal or disruption of internal architecture of the mesial temporal structures on T2-weighted or fluid-attenuated inversion recovery (FLAIR) images. All patients had undergone 0.5–1.5 T brain MRI, with T1-weighted sequences in coronal, sagittal, and axial views and additional T2-weighted or FLAIR sequences, including at least one coronal view perpendicular to the main axis of the hippocampus.

Cases fulfilling study inclusion criteria were uniformly advised to have surgery—anteromesial temporal lobectomy on the MTS side, without invasive monitoring, regardless of ipsi or contralateral temporal lobe electrographic seizure onset. In the study period, no patients with uMTS and contralateral seizure onset on vEEG monitoring underwent invasive or semiinvasive monitoring in our service. All cases with contralateral temporal lobe electrographic seizure onset were submitted to the sodium amytal test to ensure adequate memory reserve.

All surgeries were performed by the same experienced neurosurgeon (H.T.W.), with a standard surgical procedure (amygdalohippocampectomy with neocorticectomy) (Wen et al., 1999; Wen et al., 2006). The amygdala, head, and part of the body of the hippocampus were resected and sent for pathological analysis.

All patients had routine postoperative outpatient clinic follow-up visits.

Data analysis

We retrospectively collected data, through chart review and patient interview regarding age at surgery, gender, occurrence of an initial precipitating insult, age at epilepsy onset, and epilepsy duration.

vEEG monitoring

All patients underwent continuous noninvasive vEEG monitoring with at least two recorded seizures. Continuous vEEG monitoring was performed with a 64-channel BMSI-Nicolet 5000 vEEG equipment (Nicolet Biomedical, Madison, WI, U.S.A.). A modification of 10-20 System was used, as approved by the American EEG Society in 1990, in all patients. An extension of the 10-20 System was used to designate the 10% electrode placements. Positions 10% inferior to the standard frontotemporal electrodes were designated F9/F10, T9/T10, and P9/P10, and additional anterior temporal electrodes designated T1 and T2 (total of 27 electrodes) were placed to record all current seizure types. Neurophysiological data ictal and interictal EEG during noninvasive vEEG monitoring were retrospectively reviewed.

Noninvasive vEEG data: Ictal EEG

A neurophysiologist (R.M.V.), blinded to MRI and surgical outcome data, analyzed the laterality of ictal EEG tracings in all recorded seizures during vEEG monitoring. All recorded seizures were evaluated in terms of the location and side of the initial ictal EEG changes, as well as location and side of full electrographic seizure development. After comparison with MRI results by another examiner (M.H.S.), patients were classified as: (1) concordant ictal EEG, all recorded seizures originating ipsilaterally to MRI evidence of MTS; (2) discordant ictal EEG, at least one seizure originating contralaterally to MRI evidence of MTS; and (3) indeterminate ictal EEG, nonlateralizing ictal EEG in all seizures.

Discordant ictal EEG cases were further subdivided in four subgroups according to the number of contralateral ictal onset seizures, in relation to the MRI evidence of MTS: (1) subgroup one, only one contralateral ictal onset seizure; (2) subgroup two, more than one seizure with contralateral onset, but contralateral onset in less or equal than 50% of recorded seizures; (3) subgroup three, more than 50% of the seizures with contralateral onset, but not contralateral onset in all recorded seizures; and (4) subgroup four, contralateral onset in all recorded seizures.

Noninvasive vEEG data: Interictal EEG

Interictal EEG during the period of vEEG monitoring was also analyzed by the same blinded neurophysiologist. Interictal EEG activity was classified as: (1) concordant interictal EEG, when interictal epileptiform activity was seen ipsilaterally to the MRI evidence of MTS, with no more than occasional contralateral epileptiform activity (corresponding to approximately 85% or more epileptiform activity ipsilateral to MRI evidence of MTS); (2) bilateral concordant interictal EEG, when interictal epileptiform activity was seen bilaterally, although predominantely on the same side of the MRI evidence of MTS (corresponding, approximately, to more than 15% but less than 40% of contralateral epileptiform activity); and (3) discordant interictal EEG, when interictal epileptiform activity was seen bilaterally, without a side predominance or when interictal activity occurred predominantly on the opposite side of the MRI evidence of MTS.

Seizure outcome

Medical charts were reviewed and all patients were contacted with telephone calls by one independent examiner (C.L.J.), blinded to neurophysiological and MRI data. Patients were classified regarding seizure outcome according to Engel's scale (The Seizure Frequency Scoring System) (Engel et al., 1993). Engel's class I (seizure-free) was considered a good surgical outcome, whereas Engel's classes II–IV (nonseizure-free) were regarded as a poor surgical outcome. Postoperative follow-up period was considered at the moment of the last contact (telephone call or last clinic visit).

Pathological analysis

Surgical specimens were reviewed in all cases by a pathologist (S.R.), who was blinded to neurophysiological, MRI, and surgical outcome data. Hippocampal sclerosis (HS) was defined by neuronal loss in CA1, CA3, and CA4 regions of the hippocampus.

Group comparison

We compared surgical outcome in the concordant and in the discordant ictal EEG groups and in the concordant, bilateral concordant and discordant interictal EEG groups. Chi-square or Fisher's exact tests were used for statistical comparison, with a significance level of 0.05.

SPECT studies

Ictal and interictal SPECT studies (when performed) were analyzed, in that order, by visual analysis by a nuclear medicine specialist (C.R.O.), blinded to MRI, vEEG, and clinical data. Ictal SPECT was performed within less than 90 s after seizure onset during a vEEG recorded seizure, in all studies, with an intravenous (IV) injection of approximately 740 MBq (20 mCi) of 99mTc-ECD, followed by image acquisition with a dual head γ camera with dedicated collimator for brain studies (fan beam), E-CAM (Siemens, Hoffman Estate, IL, U.S.A.). Rotation time was determined through counts per frame (at least 100,000 counts per frame) and 128 projections were acquired. SPECT images were processed on a workstation E-soft (Siemens) with Butterworth filter (with 0.75 of cutoff and order 5). Reconstruction yielded 4.8 mm voxels with a 128 × 128 matrix and 128 slices. In-plane spatial resolution was 10.6/6.7 mm full-width at half maximum in the center of view.

Increased temporal lobe flow was defined as increased flow in the mesial or lateral temporal region, with or without posterior extension. Laterality was defined as the side of increased flow or the predominant increased flow side if there was bilaterally increased flow. If no lateral predominance was noted, the case was classified as bilateral. If a different flow pattern was seen, the case was classified as indeterminate.

SPECT results were then analyzed in concordant and discordant groups (defined by ictal EEG data). Since on any individual patient, seizures could originate on the ipsilateral, contralateral, or both sides in relation to the MRI evidence of MTS, we analyzed ictal EEG findings in all cases for the ictal SPECT injection seizure and then compared ictal SPECT results with MTS side.

SPECT results were classified as: (1) concordant SPECT, ictal EEG in SPECT injection seizure—MRI—when all data were ipsilateral; (2) EEG corrected by SPECT, when ictal EEG in SPECT injection seizure was contralateral to the MRI evidence of MTS, and the SPECT study showed increased flow on the MTS side; and (3) EEG not corrected by SPECT, both SPECT and ictal EEG in SPECT injection seizure were contralateral to MRI evidence of MTS.

Surgical outcome was compared in these groups.

Ethics committee approval

This research protocol was approved by the institutional review board on 07/29/2004 (protocol 591/04). All patients signed an informed consent form.


Patients selection and clinical demographic data

Ninety-four patients with uMTS underwent vEEG monitoring in the period between Jan 1999 and April 2005 in our service. Sixteen patients were not included in the study: five patients (four in the concordant and one in the discordant group) decided not to undergo surgery, seven (six in the concordant and one in the discordant group) were operated by other surgeons, three had extratemporal epileptiform activity on ictal EEG, and one had both dual pathology and psychogenic seizures recorded during vEEG monitoring. No patients were excluded because of a low IQ. Of the 78 remaining patients fulfilling inclusion and exclusion criteria, one patient in the concordant vEEG group was lost to postoperative follow-up.

Of the final 77 patients in the study (Fig. 1), forty (52%) were male. Thirty-seven (48%) had right MTS. Mean age at surgery was 34.8 ± 10.2 years (range 12–65, median, 35). Epilepsy duration was 24.2 ± 10.8 years. Thirty-three of 77 (42.8%) patients had a history of an initial precipitating insult, at a mean age of 24.2 ± 10.8 months. All patients included in the study were contacted in the last month of the study (May 2007), except for one patient who had died of an unrelated cause (suicide) 37 months after surgery. Mean postoperative follow-up was 57.3 ± 20.7 months, (range 24–94, median 63).

Figure 1.

Study design: Patient selection, classification, and surgical outcome. uMTS, unilateral mesial temporal sclerosis; vEEG, video EEG; GO, good outcome; PO, poor outcome.

Ictal EEG concordance/discordance and surgical outcome

After ictal EEG analysis, twenty-two patients (28.6%) were classified as discordant ictal EEG. One patient was classified as indeterminate ictal EEG. The remaining 54 (70.1%) patients were classified as concordant ictal EEG. Long-term follow-up period was similar in both groups (concordant: 56.1 ± 20.7 months; median 62, range 24–94 versus discordant 59.8 ± 21.2 months; median 66, range 25–88).

Concordant and discordant groups were comparable in terms of age at epilepsy onset (11.1 ± 7.5, median 10, range 1–32 years versus 10.5 ± 8.9, median 7, range 2–35), epilepsy duration (23.5 ± 10.6, median 24 range 5–47 years versus 24.8 ± 10,9, median 24.5, range 7–46 years), and presence of an initial precipitating insult (23 of 54 or 42.6% versus 10 of 22 or 45.5%), which had occurred at an age of 14.4 ± 8.8, versus 19.4 ± 12.2 months.

Surgical outcome was good in 59 of 77 (76.6%) of the cases. Surgical outcome did not differ between groups: 40 of 54 (74.1%) patients in the concordant ictal EEG group had good surgical outcome, compared with 19 of 22 (86.4%) patients in the ictal discordant group (p = 0.83). Results are shown in Table 1. The patient with indeterminate ictal EEG results had a poor surgical outcome (Engel class IIB).

Table 1.  Surgical outcome in ictal EEG concordant and discordant groups
MRI × vEEGGood outcome
  1. MRI, magnetic resonance imaging; vEEG, video EEG; n, number of good outcome cases in the group; N, number of cases in the group.

Indeterminate0/1 0.0

A detailed description of the ictal-EEG in all recorded seizures, MTS side, pathology results, surgical outcome, and follow-up length of all discordant cases is given in Table 2.

Table 2.  Ictal EEG-MRI discordant cases: seizure onset and development side for all cases
PatientMRISeizuresNo. seizuresIctal EEG OnsetDevelopmentIctal vEEG lateralizationHAOutcomeFollow-up
  1. MRI, magnetic resonance imaging; vEEG, video EEG; seizures, number of recorded seizures during video EEG monitoring; R, right; L, left; B, bilateral; R > L, bilateral, right predominating; L > R, bilateral, left predominating; G, generalized; I, indeterminate; HS, hippocampal sclerosis; NA, not analyzable.

 4R15 10 RRBHSIA68
 6R22B (L > R)RLHSIIA25
1RB (R > L)    
 9R14 6BLBHSIA83
1B (L > R)R    
3B (R > L)B (R > L)    
1B (L > R)B (L > R)    
2B (R > L)L    
1LB (L > R)    
18R15 1ILBNAIB65
3B (L > R)B (L > R)    
2RB (L > R)    
20R13 4RRBHSIA74
2BB (L > R)    

The degree of ictal discordance did not influence surgical outcome. Subgroup one (only one seizure originating contralaterally) consisted of only one patient (case eight), who displayed only one (out of six) recorded seizure originating contralaterally to the MTS side. This patient had the most unfavorable surgical outcome (Engel's IVA) in the study. In subgroup two (≤50% seizures with contralateral onset), all six cases had a good surgical outcome (cases 1, 4, 9, 12, 13, and 21; with 26 out of 57 or 45.6% seizures with contralateral onset). In subgroup three (more than 50% seizures with contralateral onset), 7 out of 8 cases (87.5%) had a good surgical outcome (cases 2, 3, 14, 16, 18, 19, and 20; with 36 out of 58 or 62.1% seizures with contralateral onset). In the most discordant group, where all recorded seizures originated contralaterally (cases 5, 6, 11, 15, 17, and 22), 5 out 6 cases (83.3%) had a good surgical outcome.

Interictal EEG concordance/discordance and surgical outcome

Interictal EEG data analysis was possible in 76 of the 77 patients. Sixty-six of these patients had concordant interictal EEG: 56 of 76 (72.7%) patients were classified in the fully concordant group, and 10 of 76 (13.1%) patients were classified in the bilateral concordant group. The remaining 10 patients (13.1%) were classified as interictal EEG discordant cases.

Long-term follow-up period was similar in the three groups (concordant: 58.2 ± 20.7 months; median 65, range 24–94 versus bilateral concordant 59.3 ± 19.4 months; median 64, range 27–88 versus discordant 52.8 ± 23.9 months; median 44.5, range 25–86).

Surgical outcome did not differ among groups. Surgical outcome was good in 52 of 66 (78.8%) in the concordant group, in 9 of 10 (90%) of the bilateral concordant group, and in 7 of 10 (70%) of the discordant group (p = 0.92).

Ictal SPECT studies and surgical outcome

Ictal SPECT was performed in 48 of 77 (62.3%) patients: 33 of 54 (61.1%) of the ictal EEG concordant cases, and 14 of 22 (63.6%) of the ictal EEG discordant cases, and in the one patient with indeterminate ictal EEG. Ictal SPECT results were analyzed separately for ictal-EEG concordant and discordant cases. A summary of ictal SPECT results, relation to MRI-ictal EEG data and surgical outcome is shown in Table 3.

Table 3.  MRI versus other tests: concordance and surgical outcome
vEEGIctal EEG during SPECTIctal SPECTGood outcome
  1. vEEG, video EEG; C, concordant test finding to MRI; D, discordant test finding to MRI; I, indeterminate test result.

C = 33C = 32C = 2721/27 (77.8%)
I = 52/3 (66.7%)
I = 1I = 11/1 (100%)
D = 14C = 6C = 43/4 (75%)
I = 22/2 (100%)
D = 8C = 53/5 (60%)
D = 33/3 (100%)

Ictal EEG-MRI concordant cases group

In this group, ictal EEG during the SPECT injection seizure was, as expected, lateralized to the MRI lesion in 32 of 33 (97.0%) cases. In one case, where other recorded seizures allowed a classification as a concordant case, ictal SPECT was injected in a seizure without an ictal EEG correlate. Ictal SPECT showed increased flow in 27 of 32 (84.4%) cases. Of these cases 21 of 27 (77.8%) had a good surgical outcome. In the remaining five cases, ictal SPECT was nonlateralized. Three out of five (60%) patients with nonlateralized SPECT in this group had a good surgical outcome. In only one patient in the concordant group, both ictal EEG during SPECT injection and ictal SPECT were nonlateralized. This patient had a good surgical outcome.

Considering the whole group of concordant cases who underwent ictal SPECT, 25 of 33 (75.6%) cases had a good surgical outcome, with a mean follow-up of 55.7 ± 21.1 months (median 62, range 24–93).

Ictal SPECT results in ictal EEG discordant cases

Ictal SPECT was performed in 14 of 22 (63.7%) discordant cases. Although all patients in this group had at least one seizure where ictal EEG was discordant to the MTS, ictal SPECT could have been injected in an ictal EEG concordant seizure. We therefore analyzed surgical outcome in three different subgroups of patients in relation to SPECT findings: (1) concordant SPECT-ictal EEG in SPECT injection seizure MRI. This group consisted of six patients. In four cases, ictal SPECT showed increased flow in the MRI-lesion side. Three out of four patients (75%) had a good surgical outcome. In the remaining two patients, ictal SPECT was nonlateralized compared to MRI. Both patients had a good surgical outcome. Mean follow-up in this group (six patients) was 71.0 ± 23.9, median 80.5, range 25–88 months. (2) EEG corrected by SPECT; discordant SPECT-ictal EEG in SPECT injection seizure MRI, but concordant SPECT-MRI. This group consisted of five patients in whom ictal SPECT showed increased flow contralaterally to the ictal-EEG, but ipsilaterally to the MRI evidence of MTS. Three out of the five cases (60%) had a good surgical outcome. Mean follow-up in this group was 57.2 ± 14.1, median 60, range 35–70 months. (3) EEG not corrected by SPECT; both SPECT and SPECT-ictal EEG in SPECT seizure were contralateral to MRI evidence of MTS. This group consisted of three patients, in whom both ictal SPECT and ictal EEG in SPECT injection seizure lateralized to the contralateral side of the lesion. All three patients in this group had a good surgical outcome; all patients were Engel's class Ia. Mean follow-up in this group was: 55.7 ± 25.1 months, median 66, range (27–74) months.

Pathological diagnosis

A pathological diagnosis was possible in 62 of 77 (80.5%) cases. In the remaining 15 of 77 (19.5%) cases, pathological material was fragmented, hindering pathological analysis. Fragmented pathological material was equally distributed in the concordant and discordant groups (10 of 54 or 18.5% versus 5 of 22 or 22.7%). HS was confirmed in 60 of 62 (96.8%) cases where pathological analysis was possible. In the two cases where pathology did not show HS, one was in the concordant group, and the other in the vEEG-MRI indeterminate group. In two other concordant cases, HS was associated with heterotopic gray matter nodules.

Good surgical outcome in the pathologically confirmed cases was similar in the concordant and discordant groups (32 of 43 or 74.4% versus 14 of 17 or 82.3%), including the two cases with heterotopic nodules (both with good surgical outcome). Of the non-HS cases, the patient with indeterminate vEEG-MRI patient had poor surgical outcome (Engel's class IIa), and the other patient (in the concordant group) had a good outcome, in a 29-month follow-up.

In the cases with fragmented material, not amenable to pathological evaluation, surgical outcome was similar in the concordant and discordant groups (7 of 10 or 70% versus 5 of 5 or 100%).


Mesial temporal lobe epilepsy (TLE) associated with HS is a surgical remediable syndrome, with 64%–93% (McIntosh et al., 2001) patients being rendered seizure-free after standard anteromesial temporal lobectomy. The MRI finding of uMTS is a potent predictor of surgical success in patients with medically refractory mesial TLE (Baulac et al., 1994; Berkovic et al., 1995). There is a high concordance of MRI, interictal and ictal EEG data in these cases (Cascino et al., 1996). In cases where presurgical evaluation indicates concordant MRI, interictal, and ictal EEG findings, surgery is indicated, with good surgical outcome, obviating invasive monitoring studies. Difficulties arise when noninvasive vEEG monitoring, and to a lesser extent, interictal EEG findings yield discordant MRI and electroencephalographic data. In approximately 3%–19% of MTS cases, noninvasive vEEG recordings show contralateral, bilateral, or ambiguous seizure onset (Williamson et al., 1993; Ebner & Hoppe, 1995; King et al., 1997). In some centers, patients with discordant ictal EEG and MRI data are submitted to chronic semiinvasive or invasive vEEG monitoring with ovale foramen, depth, or subdural electrodes. Patients with discordant vEEG and imaging data after invasive monitoring may be denied surgery for fear of poor surgical results.

Surgical results in patients with discordant findings on MRI and scalp vEEG is difficult to ascertain. Although some studies have suggested a worse surgical outcome compared to concordant cases (Jack et al., 1992; Schulz et al., 2000; Tonini et al., 2004), other studies suggest that surgical result may not be worse in this group (Hirsch et al., 1991; Baulac et al., 1994; Sirven et al., 1997; Schiller et al., 1998; Hardy et al., 2003; Mintzer et al., 2004; Jeong et al., 2005). Most studies that suggest a poorer outcome in the discordant group have included patients that had not undergone MRI, nonlesional TLE cases, TLE secondary to other etiologies, bilateral MTS, as well as cases with extratemporal seizure onset (Jack et al., 1992; Schulz et al., 2000; Tonini et al., 2004). Moreover, all studies incur in a selection bias, in which patients with discordant data after invasive monitoring could be denied surgery. Thus, surgical outcome in patients with discordant ictal EEG-MRI data cannot be precisely evaluated by these studies.

Determination of ictal EEG onset may be controversial, since subtle initial ictal changes can be subjective and examiner-dependent. In this study, we found a higher proportion of discordant cases than what is usually reported in other studies. If we had used a more stringent criterion to define discordant cases, as those where both initial ictal EEG changes and full ictal development were seen contralaterally to the MRI evidence of MTS in at least one seizure, results would not be different. Of the 13 of 22 cases in the discordant group fulfilling this more stringent definition (Table 2, cases 3, 4, 7, 10, 11, 12, 13, 16, 17, 18, 20, 21, and 22), all cases had good surgical outcome (Engel's class Ia, nine cases; class Ib, two cases; and class Ic, two cases). Our subgroup analysis further corroborated the fact that the degree of discordance, measured by the proportion of contralateral onset seizures in each patient does not appear to indicate that a particular subgroup has a worse surgical outcome. The three patients with an unfavorable surgical outcome appeared to be randomly distributed in three different discordant subgroups. Of note, the only discordant case with the most unfavorable surgical outcome (case 8, Engel's class IVa)—the other two unfavorable outcome discordant patients were classes IIa and IIc (almost without seizures)—was classified in the least discordant group, with only one seizure originating contralaterally to the MRI evidence of MTS. Also, if we consider the most discordant subgroup of patients in the ictal SPECT analysis (ictal-EEG during SPECT injection seizure and SPECT discordant from MRI lesion side), all three cases in this category had a good surgical outcome (Engel's class Ia in all cases).

In this series, the finding of bilateral concordant or even discordant interictal EEG activity did not influence outcome. Although the discordant interictal EEG subgroup had a slightly worse surgical outcome, this difference was not statistically significant.

In our study, we selected a homogeneous group of uMTS patients, without other MRI lesions or extratemporal ictal onset. Groups differed only in the presence or absence of seizure onset in contralateral temporal region on surface EEG. All patients were submitted to a standard surgical procedure, by the same surgeon, regardless of concordant or discordant noninvasive vEEG data. Although five patients decided not to undergo surgery and seven patients were operated by different surgeons, and thus not included in this study, this did not lead to a selection bias, since the proportion of concordant and discordant cases was similar in the operated and not operated groups: 2 of 12 (16.7%) discordant cases in the not operated cases, compared with 22 of 77 (28.6%) of discordant cases in the study group.

Contralateral seizure onset may represent a false lateralization, where scalp electrodes detect contralateral spread of ictal activity (“burned-out hippocampus theory”) (Mintzer et al., 2004). Conversely, this may indicate true seizure onset in the contralateral temporal region, arising from structural lesions not identified by MRI on the contralateral mesial temporal structures (Babb, 1991; Diehl & Lüders, 2000). Surgical outcome should be poorer in the discordant group in this latter scenario.

In our study, surgical outcome was similar in both groups, suggesting that false lateralization is probably the underlying scenario in most discordant cases. In this study, discordant cases had even a slightly (12%) better surgical outcome than concordant cases. Both groups were very similar in terms of demographic and clinical data, suggesting that they represent homogeneous samples of the larger population of patients with unilateral MRI evidence of MTS. We did not perform a more detailed volumetric or histopathologic analysis to evaluate if discordant cases are associated with a more severe hippocampal volume loss or cell loss.

Discordant cases may display a denser interhemispheric connection, possibly through the anterior commissure, that might contribute to a faster interhemispheric spread of ictal activity in these cases, leading to false lateralization.

With our current sample size of 77 patients, the study was only powered to detect a 37% difference in outcome between groups. In order to achieve an 80% power, a sample size of approximately 240 patients would be needed. However, the discordant EEG group had an excellent surgical outcome, well within the range of concordant uMTS cases, making a type II error less likely. Another strength of this study is that both concordant and discordant groups were followed for a mean of approximately five years, making it unlikely that good outcome might represent an initial, but not sustained surgical result over time.

Ictal SPECT studies could noninvasively solve the issue of false lateralization: ipsilaterally increased flow, despite contralateral seizure onset would suggest false lateralization. Conversely, contralaterally increased flow, discordant with the MRI evidence of MTS could suggest true contralateral seizure onset. Our data suggests that this is not the case: despite relatively low numbers, surgical outcome did not differ in cases where ictal SPECT showed increased flow ipsi or contralaterally to the lesion. In these cases, ictal SPECT may also indicate seizure spread and not seizure onset. Although false lateralization on ictal SPECT occurs less commonly than with ictal EEG, it should not be interpreted as an indicator of poor surgical outcome in these cases.

Pathology data further supports these findings, confirming MRI diagnosis in almost all cases in both groups. The finding of HS associated with heterotopic gray matter in two cases and lack of HS in another two cases in the concordant group did not modify surgical outcome of the concordant group as a whole.

The findings in this study should not be interpreted as indicative that vEEG monitoring is not useful in patients with uMTS. Nonepileptic psychogenic seizures cannot adequately ruled out without vEEG monitoring. Also, this study did not analyze surgical outcome of patients with unilateral MRI evidence of MTS who showed extratemporal seizure onset on ictal EEG.

Memory decline can be a concern in temporal lobe resection, especially in discordant cases, where bilateral hippocampal dysfunction may occur. All our discordant cases underwent sodium amytal testing to ensure adequate memory reserve in the contralateral temporal structures. Further studies should address the risk of memory decline in concordant and discordant cases.

Our findings suggest that unilateral MRI evidence of MTS with seizure onset in the contralateral temporal region have a similar surgical success rate as fully concordant uMTS cases. Potentially risky invasive or semiinvasive studies (Diehl & Lüders, 2000; Velasco et al., 2006) not only may not be indicated in these cases, but it also may lead to erroneous contraindication of a surgical procedure with a high success rate in this subgroup of mesial TLE patients with discordant ictal EEG and MRI findings.


This work was supported by FAPESP—Processes 05/56464-9 (Cinapce) and 05/50067-8.

Conflict of interest: We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines. All authors have no conflicts of interest to disclose