Mesial temporal lobe epilepsy (MTLE) is the most common form of partial epilepsy syndrome in adults, and 60% to 90% of the patients were refractory to antiepileptic drugs (AEDs)^{[1, 2]}. Such medically intractable MTLE requires surgical treatment, which can achieve a complete seizure control or significantly reduced seizure attack within 2-3 years after the surgery in about 2/3 of the patients^[3]. In patients with MTLE and concurrent hippocampal sclerosis (HS), surgical intervention was reported to result in a seizure-free time of up to two decades, and the probability of remaining completely seizure-free at 12 and 18 years was 65% and 62%, respectively^[4]. Nevertheless, the surgical outcomes of the patients vary across different epileptic centers. A metaanalysis indicates that in patients with for medically intractable MTLE, the long-term seizure freedom rate (25th, 75th percentiles) following surgery was 65% (55%, 80%)^[5], but due to the inherent limitations of the studies included (not of a randomized controlled design, inclusion of cases of temporal lobe epilepsy ^[6-7], and questionable criteria for diagnosing drug-resistant epilepsy ^[8]), the validity of this conclusion remains elusive.

The postoperative seizure outcomes are subjected to influences by the disease history, neuropsychological condition, imaging examinations, results of EEG monitoring, and surgical procedures^{[9, 10]}. Previous studies suggested that the presence of HS ^{[11, 12]}, consistency between magnetic resonance imaging (MRI) findings and interictal spikes, an early age at surgery, history of febrile seizures (FS), and absence of secondary generalized tonic seizure (sGTCS) or status epilepticus were associated with a favorable postoperative seizure control, but the impacts of these factors are still controversial. In this study, we aimed to assess the therapeutic effect of surgical interventions on chronic medically intractable MTLE and analyze the possible predictors of the surgical outcomes to provide clinical evidence for surgical treatment of MTLE.

This study was approved by the Institutional Ethics Committee of Wuhan General Hospital of PLA. We retrospectively reviewed the database of patients undergoing anterior temporal lobectomy (ATL) or selective amygdalohippocampectomy (SAH) by a single surgeon (DU Hao) at the Department of Neurosurgery, Wuhan General Hospital of PLA between September, 2011 and October, 2013. Drug-resistant epilepsy was defined as failure to achieve sustained seizure freedom after adequate trials of two tolerated and appropriately chosen and used AED schedules (whether as monotherapies or in combination)^[8]. All the neuropsychological and seizure evaluations were performed by the same team of neuropsychologists and epileptologists. The patients meeting the following criteria were included in the study: established diagnoses of drug-resistant epilepsy according to the diagnostic criteria; established diagnoses of MTLE by preoperative evaluation; the epileptogenic zone was limited to the unilateral lobe or bilateral lobes, with unilateral lobe dominance (>70%); age at surgery between 1 and 60 years (for children and adolescents, the indications could be extended according to the circumstance); absence of neuropsychiatric disorders or serious generalized disorders that affected the organs. Informed consents were obtained from all the patients or their family, who underwent the surgery on a voluntary basis with full acknowledgement of the operation risk and prognosis.

The preoperative evaluation of the patients was either noninvasive or invasive. In noninvasive evaluation, the patients were examined for detailed clinical history and neurological functions and underwent long-term video EEG monitoring, high-resolution MRI, magnetic resonance spectroscopy (MRS), fluorodeoxyglucose positron emission-computed tomography (PET-CT), and examination with comprehensive neuropsychological test battery. Invasive evaluation was carried out using invasive EEG (iEEG) monitoring.

From all cases the demographic and clinical data were collected, including gender, onset age, age at surgery, disease course, disease history, preoperative intelligence, preoperative memory function, frequency of seizures at time of surgery, seizure types, scalp-EEG lateraling, iEEG lateraling, preoperative imaging studies, surgical technique, and location of surgery. Detailed clinical history was obtained from all the patients. Semiological analysis highlighted distinct heterogeneity in most MTLE patients, and the main seizure manifestations included simple focal seizures and complex focal seizures. Simple focal seizures was characterized by such manifestations as autonomic nervous symptoms, mental symptoms and special feelings (commonly of abdominal stomach gas). Complex focal seizures often began with a sudden movement termination followed by typical gestural or oralalimentary automatisms, often with other automatism. Seizures duration was more than 1 min, and the patient was often unaware of the attacks.

High-resolution MRI was performed following the routine protocols for all the patients using a 1.5T scanner (GE). The data of coronal 3D T1-weighted (W) gradientecho imaging parallel to the brainstem were acquired, and fluid-attenuated inversion recovery (FLAIR), T2-W turbo spin-echo and T1-W inversion recovery images perpendicular to the hippocampi were obtained in addition to routine brain imaging. MRS of the bilateral hippocampi and interictal PET-CT were performed in cases without significant abnormal findings in MRI.

The 32-channel scalp-EEG recording system directed by Nicolet Corporation was used. Long-term video EEG was recorded from the scalp electrodes placed according to the standard 10-20 systems. Sphenoid electrodes were also used. The patients underwent continuous video EEG monitoring lasting 2 to 30 days until more than two habitual seizures were caught.

When non-invasive evaluations failed to identify the origin of the epileptic discharges, deep electrodes were implanted into the bilateral hippocampi via a stereotactic system. At our center, the criteria for iEEG monitoring included discrepancy between ictal video EEG and imaging findings, bilateral interictal abnormalities, and multifocal ictal findings. In patients with bitemporal spikes, a standard bilateral hippocampal depth electrode implantation was performed, and additional subdural strip or grid electrodes were used when lateral temporal lobe or other neocortex involvement was suspected. iEEG monitoring lasted several days until more than two habitual seizures were caught.

The Wechsler Intelligence Scale and Wechsler Memory Scale for children or adults were used to measure the patients' intelligence and memory function, assessed by intelligence quotient (IQ) and memory quotient (MQ), respectively. In general, an IQ lower than 90 suggested impaired intelligence, and a MQ score below 90 indicated memory dysfunction.

The surgical approach was decided according to the results of preoperative assessment. All the operations were performed by the same operator under microscope. In early cases and those refusing to have intracranial deep electrode implantation, ATL procedure was performed for resections in the dominant (3.5 to 4.0 cm) or non-dominant (up to 5 cm) temporal neocortex combined with amygdalohippocampectomy ^[13]. In ALT, slightly large temporal bone window craniotomy was chosen, and a part of the anterior temporal lobe was resected to expose the temporal angle of the lateral ventricle. When the epileptogenic zone was located in the dominant hemisphere, the resection scope was limited to 4.5 cm, and in the non-dominant hemisphere, the resection scope could reach 5.5 cm without crossing the Labble's vein.

In SAH, only the temporomesial structures were resected^[13] to avoid damage to the lateral temporal lobe cortex or impairment of lingual function, especially when the lesion was located in the dominant hemisphere^[14-15]. As resections in the dominant hemisphere was associated with greater postoperative naming ability decline^[15], the majority of the patients underwent SAH. In SAH procedure, a straight incision was made in the temporal scalp and through a keyhole approach, the lateral ventricle was accessed through the sylvian fissure (SF) or the superior temporal sulcus (STS) to expose the temporal angle of the lateral ventricle and the hippocampal structures. The parahippocampal gyrus was aspirated in a piecemeal fashion, and the hippocampus was removed en bloc. The amygdala was resected parallel with the roof of the ventricle. The resected specimens were sent for routine pathological examination, using immunohistochemical staining if necessary.

After recovery from anesthesia, the patients were given AEDs, either according to the original treatment plan or with adjustment of the type or dosage of AEDs. The majority of patients took carbamazepine or oxcarbazepine postoperatively. On the second day after surgery, the patients underwent head CT examination to ascertain that such complications as secondary bleeding did not occur in the operating field. The efficacy of the surgeries was evaluated at 3, 6, and 12 months postoperatively. The patients were followed up every year to record the conditions of seizure control, scalp EEG finding, memory function and intelligence. All the patients were asked to take AEDs for at least 2 years after the operation, and decision of dose adjustment or discontinuation of the drugs was made according to the condition of seizure control.

The annual and latest seizure outcome were described in detail according to the classification system of Engel's Classes Ⅰ- Ⅳ^[16] : Class Ⅰ, no seizures, with or without aura; Class Ⅱ, 1 or 2 seizures days per year, or at least a 90% reduction in seizures, with or without aura; Class Ⅲ, seizures are reduced by at least 50%; Class Ⅳ, no improvement of epileptic seizures.

Seizures in a few weeks after the operation (the Commission on Neurosurgery of the ILAE suggested a 4-week period) were defined as early seizures, and were not taken into account in statistical analysis of the curative effect (because the risk of a low AED level and the acute surgical outcomes like electrolyte imbalance, hypoxia, hypoglycemia or hypocapnia might be the most pronounced^[16-17]). For convenience of analysis, the patients were categorized into two groups by tailored epilepsy surgical outcomes, namely favorable outcome group (Engel's Class Ⅰ and Ⅱ) and poor response group (Class Ⅲ and Ⅳ).

All the data were analyzed using SPSS software Version 19.0. The measurement data were presented as Mean ± SE. Due to the limited number of cases, Logistic regression analysis could not be applied. Mann-Whitney U test or Kruskal-Wallis test was used to reveal the impact of the possible predictors on the surgical outcomes. A P value less than 0.05 was considered to indicate a statistically significant difference.

Among the 34 patients with medically intractable MTLE, preoperative evaluation identified 4 patients who showed characteristic electric actions derived from the bilateral mesial temporal lobe, and they did not receive surgical interventions. The demographic and clinical data and results of preoperative evaluations for the remaining 30 patients are shown in Tab. 1. No postoperative intracranial infection, personality changes, paralysis or aphasia were observed in these 30 patients during the follow up ranging from 3.5 to 5.5 years.

表 1(Table 1)

Table 1 Demographic and clinical characteristics and results of preoperative evaluation of the 30 surgical patients Categories Values Percentage Gender   Male 18 60.0%   Gemale 12 40.0% Onset age (year) 18.7±2.0 Course (year) 14.0±1.7 Age at surgery (years) 32.7±2.3 Seizure frequency at the time of surgery 13.8±4.0   3/month   3/month-3/week 10 33.3%    > 3/week 13 43.3% Seizure types 7 23.3%   CPS   sGTCS 11 36.7% History 19 63.3%   No apparent abnormality   Brain injury 22 73.3%   Encephalitis 3 10.0%   FS 3 10.0% Preoperative intellegence 2 6.7%   No apparent abnormality   Intelligence decline 27 90.0% MRI or MRS findings 3 10.0%   No apparent abnormality   With HS 7 23.3%   Lesions 16 53.3%   Days with scalp-EEG 7 23.3%   Number of seizures with scalp-EEG 5.4±0.5 Scalp-EEG-based diagnosis 3.0±0.3   Unilateral   Bilateral 10 33.3% Surgical types 20 66.7%   SAH   ALT 25 83.3% Surgical side 5 16.7%   Dominant   Non-dominant 17 56.7%   Non-dominant 13 43.3% ATL: Anterior temporal lobectomy, CPS: Complex focal seizures, FS: Febrile seizure; HS: Hippocampal sclerosis; SAH: Selective amygdalohippocampectomy; sGTCS: Secondarily generalized tonicclonic seizure.

Table 1 Demographic and clinical characteristics and results of preoperative evaluation of the 30 surgical patients

Follow-up of the patients for more than 3.5 years showed that 23 (76.7%) of the patients were in Engel's Class Ⅰ (21 patients) or Ⅱ (2 patients), and 7 (23.3%) patients were in Engel's Class Ⅲ (5 patients) or IV (2 patients).

Mann-Whitney U test or Kruskal-Wallis test showed that gender, onset age, age at surgery, course, preoperative intelligence, frequency of seizures at the time of surgery, seizure types, scalp EEG lateraling, iEEG lateraling, preoperative imaging studies, surgical technique, or location of surgery had no obvious impact on the surgical outcomes (P>0.05). The patients with no special disease history appeared to be more likely to have a favorable surgical outcome than those who experienced trauma, FS, or encephalitis (P=0.041, Mann-Whitney U test; Tab. 2).

表 2(Table 2)

Table 2 Analysis of possible predictors of surgical outcome of MTLE Possible predictors Favorable outcome group Poor response group P Gender 0.499   Male 13 5   Female 10 2 Onset age (year) 20.6±2.3 12.4±3.4 0.389   ≤10 4 3   10-20 9 2   >20 10 2 Disease course (year) 13.0±2.0 17.4±2.2 0.322   ≤10 12 2   10-20 6 4   >20 5 1 Age at surgery (year) 33.6±2.8 25.3±1.5 0.966   ≤20 4 1   20-30 9 3   30-40 2 1   >40 8 2 Seizure frequency 13.5±4.3 14.3±10.7 0.821   ≤3/month 9 2   3/month-3/week 10 4   >3/week 4 1 Seizure types 0.168   CPS 10 1   sGTCS 13 6 Disease history 0.041*   No apparent abnormality 19 3   Brain injury or encephalitis or FS 4 4 Preoperative intelligence 0.066   Normal 22 5   Intelligence decline 1 2 MRI or MRS findings 0.804   No apparent abnormality 5 2   With HS 12 4   Lesions 6 1 Scalp-EEG-based diagnosis 0.372    & #60; 70% unilateral 7 4   70%-100% unilateral 7 2   100% unilateral 9 1 Surgical types 0.976   SAH/LS 10 3   SAH/SLS 9 3   ALT 4 1 Surgical side 0.408   Dominant 14 3   Non-dominant 9 4 ATL: Anterior temporal lobectomy, CPS: Complex focal seizures, FS: Febrile seizure; HS: Hippocampal sclerosis, SAH/SF: The sylian fissure approach of selective amygdalohippocampectomy, SAH/STS: The superior temporal sulcus approach of selective amygdalohippocampectomy, sGTCS: Secondarily generalized tonic-clonic seizure. *P < 0.05, Mann-Whitney U test.

Table 2 Analysis of possible predictors of surgical outcome of MTLE

Because only 11 patients with MTLE underwent preoperative memory assessment, the effect of preoperative memory ability on the surgical outcome was analyzed in only these patients. Tab. 3 shows that 4 patients had a MQ no less than 90, suggesting a significant decline of memory function, which, however, did not obviously correlate with the surgical outcomes (P= 0.673, Mann-Whitney U test).

表 3(Table 3)

Table 3 Memory assessment of 11 patients with MTLE and their surgical outcomes Case No. Long-term memory Short-term memory Instant memory Memory quotient Outcome 1 43 51 6 100 Favorable 2 34 47 13 94 Favorable 3 38 60 10 108 Favorable 4 41 42 3 86 Favorable 5 22 45 0 67 Poor 6 40 44 0 84 Favorable 7 23 12 0 35 Favorable 8 42 51 10 103 Poor 9 46 52 15 113 Favorable 10 44 50 15 109 Favorable 11 33 54 10 97 Favorable

Table 3 Memory assessment of 11 patients with MTLE and their surgical outcomes

The special auras, such as the typical automatism and autonomic nerve symptoms, have important value in location of the epileptogenic zone in patients with MTLE^{[2, 18, 19]}. Preoperative MRI or MRS has only limited value in location of the epileptogenic zone, and only 58%- 72% of the patients with TLE showed signs of HS on MRI, with 16% of the patients with medically refractory TLE having no MRI abnormality at all^{[20, 21]}. Scalp EEG monitoring is a good noninvasive modality capable of characterizing the distribution of epilepsy activities, especially with the use of sphenoid electrode, which allows efficient identification of the spike wave. Nevertheless, scalp EEG is vulnerable to influences by impedance between the brain and electrodes and by muscle activities, and has difficulties in recording activities of the deep lobe and cortical sulci. In addition, the scope of epileptogenic zones defined by scalp EEG might be wider than it actually is. These disadvantages can be remedied by implantation of bilateral deep electrodes, which are capable of recording subtle activities of the bilateral hippocampi without interference signal to allow accurate positioning of the epileptogenic zones. In this study, we found that implantation of bilateral deep electrodes was necessary in patients with typical simple focal seizures with little or no sGTCS, while cortical electrodes was essential for those who showed automatism early with sGTCS to confirm whether the neocortex was affected by seizures.

The complications of temporal lobe epilepsy surgery contain included transient or persistent hemiplegia, quadrantanopia, aphasia and cognitive dysfunction, and personality changes. The patients in this study did not show any signs of postoperative intracranial infection, personality changes, paralysis or aphasia. Favorable outcomes were obtained in 76.7% of the cases with MTLE, and 23.3% had a poor response to the surgeries. A long-term follow-up study of 434 patients with TLE showed that surgical interventions resulted in a high rate (69.4%) of favorable surgical outcome (Engel's class Ⅰ) at as long as 16 years after the surgeries^[7]. Although the postoperative remission rate of TLE vary across different reports ^[9], they all suggested that the rate tended to decrease year by year. But due to the inclusion of also TLE cases secondary to obvious lesions of the temporal lobe, the conclusions of these studies need to be carefully weighed before an accurate analysis can be made of the long-term surgical outcomes of patients with chronic MTLE.

In our cases, 6 out of the 7 cases with poor prognosis received frequent sGTCS (5-12 times per year) with typical automatism and typical aura of medial temporal lobe epilepsy; 1 patient showed obviously impaired intelligence. For these patients, it can be crucial to understand whether the epileptogenic zones was beyond the routine resection area or was located outside the medial temporal lobe. The possibility should be also considered that a new ipsilateral epileptogenic zone had occurred as a result of long-term epileptic seizures.

We showed that the surgical outcomes of the patients had no obvious correlation with the patients' gender, age of onset, age, epilepsy surgery, preoperative seizure frequency and duration, seizure types, imaging findings, preoperative mental status, preoperative scalp-EEG, operation method, or the operation side, as consistent with the findings in other studies^{[5, 22, 23]}. We performed Wechsler memory test in only 11 patients, and the memory function of the patients did not seem to affect the surgical outcomes; 3 patients were found to have impaired intelligence before the operation, but in this study we could not assess the correlation between intellectual damage and the surgical outcomes.

We found that patients with no special disease history (trauma, febrile seizure, or encephalitis) tended to have a more favorable surgical outcome. Currently the causal relationship between HS and TLE has not been established; patients with HS may not have seizures, and TLE patients may not have HS^[24]. We found only one patient with MTLE that had a family history of epilepsy, but the association between a family history of epilepsy and surgical outcome could not be established.

Unilateral discharge in scalp EEG is thought to be associated with a favorable surgical outcome^[25], whereas bilateral synchronous discharge or generalization from one side to the other side can be associated with a poor curative effect^[26]. In this study, unilateral localization in scalp EEG monitoring could be seen only in 30% MTLE patients, and most of the patients had interictal bilateral discharges; statistical analysis showed that neither unilateral nor bilateral discharges were significantly correlated with the surgical outcomes. The discrepancy between our finding and previous studies may result from a relative small sample in the study and the differences in the case selection. In this study, we considered only nonlesional MTLE and chronic secondary lesional MTLE. Many previous studies may not have been as stringent in differentiating MTLE and neocortical TLE, and some only examined MTLE cases with HS ^[25-27].

ATL is a well-established and effective surgical procedure for treatment of intractable MTLE. However, in some MTLE patients, ATL might result in worsening of cognitive functions, particularly lingual and memory functions, when epileptic discharges originate from the dominant hemisphere ^{[14, 28]}. SAH might reduce those damages by sparing the nonepileptogenic structures of the neocortex^[15]. Recent meta-analyses suggested that compared with SAH, ATL was likely associated with a reduced rate of seizure recurrence^{[27, 29]}, while several studies suggested that SAH was associated with better cognitive outcomes^{[27, 30]}, and ATL could result in memory deficits, particularly verbal memory following dominant hemisphere resection^[31]. Considering the advan-tages of the SAH over ALT in terms of brain protection, we recommend SAH via the superior temporal sulcus approach as the optimal surgical procedure for treatment of MTLE.