Epilepsy is a common disorder with an incidence of 50/100,000 per year and a prevalence of 5–10/1000 in North America [
Despite improved outcome, seizure recurrence after resective surgery is not uncommon. A multicenter study demonstrated that only 50–68% of patients remained completely seizure-free after anterior temporal lobectomy (ATL) with two or more years of follow-up [
From the epilepsy database, we retrospectively identified the patients who underwent resective surgery between January 2006 and July 2012 at Parkland Memorial Hospital, affiliated with the comprehensive epilepsy program at the University of Texas Southwestern Medical Center. All patients had inpatient video-EEG monitoring and brain MRI. Positron emission tomography (PET), single-photon emission computed tomography (SPECT), neuropsychological assessment, and Wada test were performed in selected patients depending on clinical indication. Invasive monitoring with subdural grid electrodes or depth electrodes was performed when the scalp EEG findings were inconclusive. Intraoperative electrocorticography (ECoG) was done in selected patients to further tailor the resection area. The patients were discussed in the multidisciplinary epilepsy conference to reach consensus before proceeding with the surgery. The resected tissue was evaluated by experienced neuropathologists. We included patients who underwent surgery for intractable epilepsy during the above period and had at least 2 years of postoperative follow-up in our center. We excluded patients whose postoperative pathology confirmed high-grade malignant tumor because their outcome could be significantly influenced by the underlying tumor itself.
From chart review, we extracted the following demographic and clinical data: age, gender, preoperative seizure frequency, epilepsy duration prior to surgery, history of secondary generalized tonic-clonic seizures (SGTCS), number of antiepileptic drugs (AEDs) tried prior to surgery, resection site, ictal EEG findings, interictal EEG findings, imaging findings, Wada memory lateralization, and pathological findings. We did not consider auras in estimating the seizure frequency. MRI was considered as abnormal only if the observed findings were consistent with well-established, potentially epileptogenic entities; in other words, isolated abnormalities that are unlikely to cause seizures (e.g., chronic microvascular disease and nonspecific white matter changes) were not considered as abnormal. We classified the resection site as temporal or extratemporal. We determined the ictal onsets based on established criteria [
We reviewed the charts of patients who had at least 2 years of postoperative follow-up. Seizure outcome was assessed using Engel classification [
Longitudinal outcome was evaluated at annual intervals. Outcome at the 2-year interval was classified as class I if the patients remained seizure-free for the 2-year period prior to the follow-up visit. Starting at the 3rd postoperative year, the outcome was classified as class I if the patients remained seizure-free for the 1-year period prior to the follow-up visit. The time to first postoperative seizure was evaluated (see below). Immediate postoperative seizures, within 1 month after surgery, were not included in the analysis.
The data range and median values were summarized for the continuous variables such as age, number of AEDs, preoperative seizure frequency, and duration of epilepsy. Continuous variables were converted into categorical variables by grouping the values into categories for univariate analysis using chi-square or Fisher’s exact tests as appropriate. Variables with
There were 78 patients eligible for inclusion in the study. Of these, 8 patients were excluded because the postoperative pathology was consistent with high-grade malignant tumor. Thus, 70 patients (32 males and 38 females) were available for analysis (Table
Demographic and clinical characteristics of the cohort (70 patients).
Gender | Male 32 (46%); female 38 (54%) |
Mean age, years (range) | 39 (21–64) |
Mean epilepsy duration, years (range) | 18 (1–57) |
History of secondary generalized tonic-clonic seizures, |
46 (66%) |
Mean number of antiepileptic drugs tried (range) | 5 (1–10) |
Mean follow-up, months (range) | 48 (24–87) |
Resection site, |
Temporal, 54 (77%); extratemporal, 16 (23%) |
Number of extraoperative invasive monitoring, |
12 (17%) |
Number of intraoperative electrocorticography, |
41 (59%) |
Ictal EEG, |
Concordant, 52 (74%); discordant, 16 (23%); inconclusive, 2 (3%) |
Interictal EEG, |
Concordant, 37 (53%); discordant, 25 (36%); normal, 8 (11%) |
MRI, |
Abnormal 53 (76%); normal 17 (24%) |
Number of patients who had Wada test, |
52 (74%) |
Wada memory lateralization, |
Concordant 44 (85%); discordant 8 (15%) |
Pathology, |
Mesial temporal sclerosis or gliosis 45 (64%); benign tumor 7 (10%); vascular lesion 4 (4%); other 7 (10%); normal 7 (10%) |
The follow-up period for assessment of seizure recurrence ranged from 24 to 87 months (mean 48.1; median 43.5 months). At the last follow-up visit, the outcome was class I in 59 (84%) patients (temporal,
During the follow-up period, 22 patients experienced seizure recurrence. Using a 2-year cut-off period, we found that the seizure recurrence followed two patterns that were clearly different (
Initial seizure recurrence pattern and subsequent outcome.
Early recurrence (≤2 years) | Late recurrence (>2 years) |
| |
---|---|---|---|
Total number of patients, |
18 | 4 | <0.05 |
Number of patients with continued seizures at subsequent follow-up, |
15 | 1 | |
Number of patients seizure-free at subsequent follow-up, |
3 | 3 |
We analyzed the postoperative outcome (seizure-free versus seizure recurrence) at various follow-up periods using univariate analysis (Table
Predictors of seizure recurrence over 5 years of follow-up.
Predictor | Number of patients analyzed ( |
|||
---|---|---|---|---|
2 years | 3 years | 4 years | 5 years | |
Age (<30 versus ≥30 years) | 70 (0.11) | 50 (0.12) | 36 (0.2) | 27 (0.63) |
Gender | 70 (0.39) | 50 (0.07) | 36 (0.24) | 27 (0.68) |
History of GTC seizures | 70 (0.08) | 50 (0.73) | 36 (0.2) | 27 (0.05) |
Epilepsy duration (<10 versus ≥10 y) | 69 (0.13) | 49 (0.07) | 35 (1) | 26 (1) |
Seizure frequency (<10/m versus ≥10/m) | 63 (0.5) | 44 (0.47) | 35 (0.51) | 26 (0.63) |
Number of AEDs (<5 versus ≥5) | 68 (0.24) | 49 (1) | 35 (0.56) | 26 (0.64) |
MRI (normal versus abnormal) | 70 (0.21) | 50 (0.17) | 36 (0.13) | 27 (0.39) |
Interictal EEG (concordant versus discordant) | 62 (0.17) | 43 (0.2) | 31 (0.46) | 25 (0.67) |
Ictal EEG (concordant versus discordant with resection) |
|
|
|
|
Resection (temporal versus extratemporal) |
|
|
|
27 (0.13) |
Wada memory lateralization (contralateral versus other) | 52 (0.27) | 38 (1) | 27 (0.17) | 19 (0.42) |
Pathology (MTS, gliosis, tumor, vascular, and other) | 70 (0.6) | 50 (0.38) | 36 (0.09) | 27 (0.42) |
AED: antiepileptic drug; MTS: mesial temporal sclerosis; GTC: secondary generalized tonic-clonic seizures.
Upon multivariate analysis using Cox proportional analysis, only two variables, resection site and ictal EEG, still retained their significance as independent predictors of seizure recurrence (Table
Predictors of seizure recurrence: multivariate analysis.
Risk ratio | 95% CI |
|
|
---|---|---|---|
Resection (temporal versus extratemporal) | 4.2 | 1.5–11 | <0.01 |
Ictal EEG (concordant versus discordant) | 5.6 | 2.0–15.7 | <0.01 |
CI: confidence interval.
Kaplan-Meier survival analysis demonstrated statistically significant differences in seizure outcome with regard to resection site and ictal EEG findings (Figure
Kaplan-Meier survival analysis of class I seizure outcome: (a) shows the comparison between temporal resection (solid line) and extratemporal resection (dashed line); (b) shows the comparison between concordant ictal EEG (solid line) and discordant ictal EEG (dashed line) with respect to the resection site.
In this study, we present our single-center experience of longitudinal seizure outcome after epilepsy surgery in a heterogeneous group of 70 patients regardless of resection site or presumptive etiology. The main findings were as follows: (1) >80% of the patients experienced class I outcome at the last mean follow-up of 4 years; (2) in patients with seizure recurrence, the majority of recurrences (>80%) occurred early (within 2 years after surgery) and a majority of such patients (>80%) continued to have seizures over the subsequent follow-up period despite medical management; and (3) among multiple variables, extratemporal resection (versus temporal resection) and discordance between ictal EEG and resection area (versus concordance between the two) predicted 4.2-fold and 5.6-fold higher risk of seizure recurrence over time, respectively.
In our group of patients who had both temporal and extratemporal resections, class I outcome was achieved in 84% at the last follow-up period (mean 48 months). These results are similar to the previous studies and meta-analysis [
Analysis of seizure recurrence patterns in our study showed that the majority of seizure recurrence (82%) occurred within 2 years after surgery, which we chose as the cut-off for “early” recurrence. This early recurrence predicted poor long-term outcome in our study, with a majority (83%) of such patients continuing to have seizures despite optimum medical management. Previous studies of temporal lobectomy demonstrated an initial phase of steep seizure recurrence at about 1-2 years, followed by a relapse rate of 2–5% per year for 5 years before stable seizure freedom was achieved [
Studies of predictors of postoperative seizure recurrence are helpful in selecting the best surgical candidates. Extensive research regarding the predictors of postoperative outcome has been done, and multiple positive or negative predictors have been proposed [
In our study, the resection site was a powerful predictor of seizure recurrence, with extratemporal resections carrying nearly a 4.2-fold higher risk of seizure recurrence than temporal resections. These results are in keeping with prior studies, which showed seizure freedom in the range of 60–70% and 30–50% after temporal lobectomy and extratemporal resections, respectively [
In our study, the ictal EEG findings concordant with the resection site predicted favorable outcome over the long term. At last follow-up, 92% of patients (48/52) with concordant EEG and 56% of patients (9/16 patients) with discordant EEG were seizure-free. This is along the lines of prior studies showing the value of EEG as a predictor of postoperative seizure outcome in the presence or absence of MRI abnormalities in patients with temporal or neocortical epilepsies [
In our study, interictal EEG was not a predictor of seizure recurrence, which is along the lines of the conclusions from a prior meta-analysis [
Our results are in agreement with other studies that have demonstrated that nonlesional MRI can be associated with an outcome as good as lesional MRI provided the scalp EEG findings are concordant with other functional studies and the planned resection site [
Besides interictal EEG and MRI findings, the other nonpredictors of seizure recurrence in our study were age, gender, history of SGTC seizures, epilepsy duration, preoperative seizure frequency, number of AEDs, Wada memory lateralization, and lesion pathology. It is well established that gender is not predictive of outcome, but the literature is conflicting as to whether the other characteristics have predictive value [
Our study has a few limitations. Because of its retrospective nature, we were unable to determine if there were discrepancies in selecting patients for surgery. However, all the patients were discussed in a multidisciplinary conference, which ensured at least some degree of uniformity. The number of patients completing the long-term follow-up beyond 5 years was smaller due to loss to follow-up. We were unable to investigate the mechanisms of seizure recurrence because only a few patients with postoperative seizure recurrence underwent follow-up EEG or video-EEG evaluation. We were also unable to ascertain how many patients remained seizure- and aura-free because of inconsistencies in documentation. Prospective studies in larger cohorts are needed for a better understanding of the pathogenesis of seizure recurrence after epilepsy surgery. Nevertheless, our study demonstrates that epilepsy surgery is beneficial in intractable epilepsy and that temporal resection and concordant ictal EEG are the major determinants of favorable outcome over long-term follow-up.
None of the authors has any conflict of interests to disclose.