Article Text

Download PDFPDF

Temporal lobe epilepsy surgery: different surgical strategies after a non-invasive diagnostic protocol
  1. P P Quarato1,
  2. G Di Gennaro1,
  3. A Mascia1,
  4. L G Grammaldo1,
  5. G N Meldolesi1,
  6. A Picardi3,
  7. T Giampà1,
  8. C Falco1,
  9. F Sebastiano1,
  10. P Onorati2,
  11. M Manfredi1,
  12. G Cantore1,
  13. V Esposito1
  1. 1Department of Neurological Sciences, Epilepsy Surgery Unit, IRCCS Neuromed, Pozzilli (IS), Italy
  2. 2Department of Human Physiology and Pharmacology, University “La Sapienza”, Rome, Italy
  3. 3Centre of Epidemiology and Health Surveillance and Promotion, Italian National Institute of Health, Rome, Italy
  1. Correspondence to:
 Dr P P Quarato
 *Epilepsy Surgery Unit, Department of Neuroscience, IRCCS “NEUROMED”, 86077 Pozzilli (IS), Italy;


Aim: To test a non-invasive presurgical protocol for temporal lobe epilepsy (TLE) based on “anatomo–electro–clinical correlations”.

Methods: All consecutive patients with suspected TLE and seizure history <2 years were entered into the protocol, which included video-electroencephalographic (EEG) monitoring and magnetic resonance imaging (MRI). Three different TLE subsyndromes (mesial, lateral, mesiolateral) were identified by combined anatomical, electrical, and clinical criteria. “Tailored” surgery for each subsyndrome was offered. Patients with seizure history <2 years, MRI evidence of temporal mass lesion, and concordant interictal EEG and clinical data bypassed video-EEG monitoring and were directly scheduled for surgery.

Results: Lesionectomy was performed without video-EEG recording in 11 patients with tumorous TLE. Of 146 patients studied with video-EEG, 133 received a TLE diagnosis. Four were excluded for neuropsychological risks, eight refused surgery, and 121 underwent surgery. Of 132 consecutive patients who underwent surgery, 101 had at least one year of follow up. They were divided into a “hippocampal sclerosis/cryptogenic” group (n = 57) and a “tumours/cortical organisation disorders” group (n = 44). In the first group, extensive temporal lobectomy (ETL) was performed in 40 patients, anteromesial temporal lobectomy (AMTL) in 17 patients. At follow up, 47 patients were seizure free. In the second group, lesionectomy plus ETL was performed in 23 patients, lesionectomy plus AMTL in six patients, and lesionectomy alone in 15 patients. Thirty nine patients were seizure free.

Conclusions: These findings suggest that different TLE subsyndromes can be identified accurately using non-invasive anatomo–electro–clinical data and can be treated effectively and safely with tailored surgery.

  • AED, antiepileptic drugs
  • AMTL, anteromesial temporal lobectomy
  • CI, confidence interval
  • DNET, dysembrioneuroepithelioma
  • EEG, electroencephalographic
  • ETL, extensive temporal lobectomy
  • ETLE, invasive presurgical evaluation
  • MRI, magnetic resonance imaging
  • MTS, mesial temporal sclerosis
  • Prot, protocol
  • TLE, temporal lobe epilepsy

Statistics from

Drug resistant focal epilepsy is responsible for high social and economic costs in industrialised countries.1

Approximately 60% of all patients with epilepsy suffer from focal epilepsy, and in one third of these patients, most of them affected by temporal lobe epilepsy (TLE), seizures are not adequately controlled with antiepileptic drugs (AED).2 Surgery is currently accepted as an effective and safe therapeutic approach in drug resistant epilepsy, particularly in TLE, where patients become “seizure free” in 70–90% of cases.3 However, surgery for epilepsy remains an underused, “last resort” treatment because only a small proportion of patients affected by surgically remediable epilepsies undergo this intervention. In a recent editorial,4 Engel reported fear of morbidity and confidence in new AEDs or in vagal nerve stimulation as factors that discourage patients and their physicians from surgery, which lacks sufficient data from randomised controlled trials. However, a recent randomised controlled trial of TLE surgery found it to be superior to prolonged medical treatment in terms of efficacy and safety.5 Moreover, although today there is agreement that TLE can be diagnosed in most patients without invasive tests,6–8 there is no consensus between different epilepsy surgery groups regarding the optimal use of non-invasive procedures.

The aim of an ideal non-invasive presurgical protocol in TLE surgery should be to: (1) identify the epileptogenic zone and consequently to identify potential candidates for intracranial investigations—that is, some patients affected by lateral TLE; (2) minimise the cost of human and technological resources; (3) reduce the time of diagnostic evaluation, so that surgery can be offered to more patients. In our centre, we implemented a non-invasive presurgical protocol for TLE diagnosis (TLE-Prot), based on “anatomo–electro–clinical correlations”, inspired by methodological principles first proposed by Bancaud and Talairach,9 and later developed by their successors.10–12 This protocol allows different types of tailored resection to be offered for TLE subtypes.13

We report the results of the presurgical evaluation and the outcome of epilepsy surgery in 101 consecutive patients with TLE, diagnosed by our protocol, who have been followed for at least one year after surgery.


Patient population

Between September 1999 and January 2003, 101 consecutive patients with drug resistant TLE who underwent surgery and who had a follow up of at least one year were entered into our study. They were from a series of patients (331 patients: 184 males, 147 females; mean age, 30.3 years; SD, 10.3; range, 6–62; mean age of epilepsy onset, 13.1 years; mean epilepsy duration, 16.2; SD, 10.2; range, 1–45) with medically refractory partial epilepsy for more than one year, referred to the epilepsy surgery unit of IRCCS NEUROMED, Pozzilli, Italy. All operated patients had undergone adequate trials of at least two first line and two add on drugs from among the new AEDs.

Presurgical diagnostic protocol

In our unit, the presurgical evaluation consists of a multilevel and individualised investigation characterised by non-invasive and invasive procedures. The invasive procedures (ETLE-Prot) are not the focus of our present study and will not be discussed.

First diagnostic level

All patients referred for surgery had a one day admission to the hospital to collect medical history data and perform preliminary neuroradiological and electroencephalographic (EEG) studies.

Medical history

A detailed medical history was obtained from all patients as a first step. Particular attention was paid to recognising the symptoms/signs at seizure onset and auras, because it is largely accepted in the literature that the initial seizure semiology usually provides valuable information about the seizure onset zone.14–16 A full clinical general and neurological examination was carried out in all patients.

EEG monitoring

The patients were monitored for 24 hours. Three 30 minute interictal EEG “standard” samples in the awake state, including hyperventilation and photic stimulation, and all sleep recordings were evaluated to assess the presence of background abnormalities, interictal slow activity, and epileptiform activity (focal (over one to three channels), regional (over three or more channels), hemispheric (over all channels of one side), and diffuse (over all channels of both sides)).17

Neuroradiological evaluation

The patients underwent brain 1.5 Tesla magnetic resonance imaging (MRI) examination and, in selected cases, brain computed tomography scans. The MRI scans were reviewed by a neuroradiologist, experienced in the field of epilepsy, masked to the clinical and outcome data. At review, MRI data were classified as follows: mesial temporal sclerosis (MTS) or lesion (low grade tumours, dysembrioneuroepithelioma (DNET), other tumour, dysplasia, or other). The presence of MTS was evaluated qualitatively by visual inspection of MRI: atrophy (on T1 weighted sequences) and increased mesial temporal signal intensity (on T2 weighted and fluid attenuated inversion recovery sequences) were considered markers of MTS. Volumetry was not performed.18,19

Second diagnostic level

The patients admitted to the second diagnostic level were evaluated by longterm video-EEG monitoring. They also underwent a comprehensive neuropsychological and psychiatric assessment, the results of which are beyond the scope of this paper and will be reported elsewhere.

The video-EEG recording technique20 was performed with collodion fixed scalp electrodes (16 EEG channels, positioned according to international 10–20 system, and one electrocardiographic channel to monitor ictal heart rate findings).21 All patients admitted to video-EEG monitoring had at least one seizure recorded.

Ictal EEG findings

We categorised ictal events as seizures with subjective phenomena only (auras) and seizures with symptoms and/or signs, with or without loss of contact (simple partial or complex partial seizures).

Each ictal event was correlated with the corresponding EEG changes, when present.

According to the site, we distinguished between focal, regional, hemispheric, or diffuse discharges.

According to the EEG seizure pattern appearance and course, ictal EEG changes were classified into ictal onset pattern (first sudden change of frequency with attenuation or appearance of a new rhythm), ictal core pattern, late patterns, and post ictal pattern.

In patients with TLE, we used a classification of EEG ictal core patterns according to well known studies22–24 concerning ictal scalp–intracranial recordings, which showed a strict correlation between ictal scalp morphology of discharge and different temporal lobe structures involved at seizure onset, as follows:

  • Type 1: antero–temporal 5–9 Hz discharge, associated with a highly probable onset in the mesial temporal structures.

  • Type 2: temporal 2–4 Hz discharge, associated with a highly probable onset in the lateral temporal structures.

EEG ictal onset in extratemporal regions excluded the patients from the TLE-Prot.

Ictal clinical findings

In selecting the patients with TLE, ictal onset semiology characterised by focal motor, sensory, visual phenomena, or complex motor manifestations (forced head turning or hypermotor behaviour) suggesting a non-temporal lobe seizure onset, led to exclusion from the TLE-Prot.

In implementing the TLE diagnostic grid, despite the fact that no ictal behaviour is specific for temporal lobe seizures, we considered some initial symptoms/signs as highly suggestive of temporal seizures and classified them into a mesial or lateral cluster (table 1).

Table 1

 Grid of anatomical, EEG, and clinical criteria used for the identification of TLE subsyndromes

TLE protocol (fig 1)

Figure 1

 Decisional algorithm for the diagnosis of TLE and surgical strategy. AMTL, anteromesial temporal lobectomy; Atr, atrophy; Dys, dysplasia; ET, extratemporal; ETL, extensive temporal lobectomy; HS, hippocampal sclerosis; I/Ictal, interictal; INV, invasive investigations; L, lateral; LES, lesionectomy; M, mesial; NPSY, neuropsycological assessment; TLE, temporal lobe epilepsy.

The first diagnostic step in selecting patients to admit to the TLE-Prot was the evaluation of the MRI examination. If MRI showed an extratemporal lesion the patient was scheduled for the ETLE-Prot, whereas in all other cases the patient was admitted to the TLE-Prot (if other investigations performed in the first diagnostic level consistently pointed to TLE).

  • If MRI showed a temporal tumour and seizure history was one to two years, the proposal of a lesionectomy was evaluated without admission to level 2 diagnostic procedures: when the tumour involved mesiolateral structures and interictal EEG was concordant, and when the tumour involved mesial or lateral structures only and both interictal EEG and the patient and eyewitness history clinical data were concordant, a diagnosis of TLE was made and lesionectomy was offered (fast TLE-Prot); if interictal EEG and, in mesial or lateral tumours, clinical data were discordant, the patients were scheduled for the second diagnostic level.

  • If MRI showed a temporal tumour and the seizure history was longer than two years, and in all other cases but extratemporal lesions, the patient was admitted to level 2 diagnostic procedures and underwent video-EEG. If both interictal and ictal electro–clinical data revealed a diagnosis of TLE, surgery was offered and the operation was individualised according to the scheme for the identification of individualised TLE surgery. If interictal EEG and/or ictal electroclinical data were discordant, the patient was scheduled for the ETLE-Prot.

  • When neuropsychological data disclosed memory deficits contralateral to the side of the epileptogenic zone, suggesting a high risk for postoperative amnesia after temporal lobe surgery, the patient was scheduled for the Wada test or excluded from surgery.

    Of 331 patients who underwent the preliminary assessment, 157 were scheduled for the TLE-Prot. MRI evaluation of these patients showed 84 hippocampal scleroses, 56 tumours (48 low grade neoplasms, eight cavernomas), 11 findings suggesting dysplasia (in four cases a hippocampal sclerosis was also evident), and four focal neocortical atrophies. In two patients, MRI showed no abnormalities. Among the 56 patients with MRI evidence of a tumour, 14 patients had a history of epilepsy for one to two years. In this group, in eight patients the lesion was located in the mesial structures and in six patients in the lateral aspects. No patient in this group had a lesion involving both the mesial and lateral structures. Three of 14 patients were excluded from a fast surgical protocol and were admitted to the second level because of non-concordance of EEG interictal or clinical data with MRI. In the 143 patients admitted to video-EEG monitoring, 929 seizures (mean number of seizures for each patient, 6.5; mean duration of monitoring days, 5.6) were recorded.

    One hundred and fifty nine patients were admitted to the ETLE-Prot and 15 patients were excluded from surgery.

    Scheme for the identification of individualised TLE surgery

    To propose individualised and limited surgery strictly according to the epileptogenic zone, we developed a diagnostic grid (table 1) based on anatomical and ictal electroclinical criteria considered to be suggestive of epileptogenesis in the mesial or lateral temporal lobe. Patients were subdivided into the following three groups:

    1. Mesial TLE: when the anatomical criterion for mesial cluster, the EEG criterion for mesial cluster, and at least two of the clinical criteria for mesial cluster were met.

    2. Lateral TLE: when the anatomical criterion for lateral cluster, the EEG criterion for lateral cluster, and at least two of the clinical criteria for lateral cluster were met.

    3. Mesiolateral TLE: when at least one criterion for mesial cluster and at least one criterion for lateral cluster were met, or when the anatomical and the EEG criteria were concordant with one clinical criterion alone.

    If brain imaging was negative, only the presence of EEG and clinical criteria were taken into account for the classification.

    This scheme was used to define different surgical strategies: lesionectomy (fast TLE-Prot, tumorous lateral TLE), anteromesial temporal lobectomy (AMTL) in mesial-TLE, and extensive temporal lobectomy (ETL) in hippocampal sclerosis/cryptogenic or lesional mesiolateral TLE.


    All operations were performed by the same epilepsy surgeon (VE).

    Both ETL and AMTL included microsurgical resection of the amygdala and en bloc excision of the hippocampal formation and parahippocampal gyrus. These interventions differed in the extent of the neocortical resection. Non-dominant ETL included excision of 4–4.5 cm of the superior temporal gyrus and the middle temporal gyrus, and 5–6 cm of the inferior temporal gyrus, whereas dominant ETL included excision of 4–5 cm of the middle and inferior temporal gyrus, although the superior gyrus was left intact. In AMTL, the extent of the neocortical excision was 3 cm for all the first three temporal gyri (sparing the superior gyrus in the dominant hemisphere). Lesionectomy consisted of complete removal of the foreign epileptogenic tissue alone.

    Seizure outcome assessment

    Seizure outcome was determined by the patient’s report to the neurologist during the scheduled follow up visits, including 60 minutes awake EEG standard recordings,25 and it was classified according to Engel.26 During the first year of follow up, AEDs were kept constant in all patients.

    Statistical analysis

    Descriptive statistics were used to summarise the data for all variables. The confidence intervals (CI) for the proportion of seizure free patients in each group were calculated. The χ2 test was used to test for differences in outcome between patients with foreign tissue lesions and patients with hippocampal sclerosis.


    A lesionectomy was offered to 11 patients with TLE after the fast TLE-Prot. After video-EEG monitoring, another 133 patients were diagnosed as having TLE, and 13 were admitted to the ETLE-Prot. Four of the 133 patients with TLE were excluded from surgery for psychiatric or neuropsychological disturbances. Therefore, surgery was offered to a total of 140 patients with TLE. Eight of these patients refused surgery.

    Of the 132 consecutive patients with TLE who underwent surgery, 101 had a follow up of at least one year (range, 12–60 months; mean duration, 30). No patients were lost to follow up, although 31 patients had a follow up period of less then one year and therefore were excluded from our study. All patients with TLE were divided in a “hippocampal sclerosis/cryptogenic” group (57 patients) (table 2) and a “tumours/cortical organisation disorders” group (44 patients) (table 3).

    Table 2

     Results of investigations and postoperative outcome in 57 patients with hippocampal sclerosis

    Table 3

     Results of investigations and postoperative outcome in 44 patients with a foreign tissue lesion

    Hippocampal sclerosis/cryptogenic group

    General characteristics

    Fifty seven patients (25 females and 32 males; mean age, 33.9 years; SD, 10.2; range, 12–62; mean age of epilepsy onset, 12.4 years; SD, 8.7; range, 4–43; mean epilepsy duration, 21.5 years; SD, 9.5; range, 2–41) were classified into this group. Risk factors for epilepsy were found in 44 patients (febrile convulsions in 35 and perinatal asphyxia in nine patients).

    Table 4 shows the results of presurgical investigations, type of operation and histopathology.

    Table 4

     Global results of presurgical evaluation, type of surgery, and histopathology in the 101 operated patients with follow up after surgery >1 year

    Seizure outcome

    Forty seven (95% CI, 70.1 to 91.2) patients were classified as Engel class I (42 as class Ia, five as Engel class Ib), seven as Engel class II, two as Engel class III, and one as Engel class IV.

    Tumours/cortical organisation disorders group

    General characteristics

    Forty four patients (21 females and 23 males; mean age, 32.3 years; SD, 12.7; range, 76–63; mean age of epilepsy onset, 19.7 years; SD, 14.7; range, 1–57 years; mean epilepsy duration, 12.5 years; SD, 11.9; range, 1–45) were classified into this group. Risk factors for epilepsy were found in eight patients (febrile convulsions in five and perinatal asphyxia in three patients). Pathology showed 21 low grade tumours, eight DNETs, five cavernous angiomas, nine cortical dysplasias, and one focal Rasmussen encephalitis.

    Table 4 shows the results of the presurgical investigations, the type of surgery, and the histopathology.

    Seizure outcome

    Thirty nine (95% CI, 75.4 to 96.2) patients were classified as Engel class I (36 as Engel class Ia, three as Engel class Ib) and five as Engel class II.


    Permanent postoperative complications occurred in two patients (2%): one patient developed a hemiplegia after ETL, whereas the other developed a hemianopia after lateral lesionectomy.

    With regard to the neuropsychological outcome, no patients suffered from serious or clinically evident neuropsychological morbidity at the one year follow up assessment.

    The seizure outcome of patients with foreign tissue lesions (39 in Engel class 1) was not significantly different from that of patients with hippocampal sclerosis (47 in Engel class 1) (χ2  =  0.75; p  =  0.39).


    TLE is the “model” surgically remediable epileptic syndrome, first because it is the most frequent among focal epilepsies, and second because it is often resistant to medical treatment. Moreover, because TLE often reflects epileptogenesis localised in “discrete” and anatomically well circumscribed structures (the amygdalo–hippocampal complex), whose radiological and electroclinical seizure correlates are well known, it may be considered particularly eligible for surgical treatment.5,6,27 Therefore, almost all centres offer surgery for TLE after a non-invasive diagnostic investigation, reporting good seizure outcome and surgery safety. However, common guidelines regarding the relative weight of the different presurgical investigations are still not well established. As a consequence, several protocols emphasise neuroimaging (MRI),28 whereas other protocols emphasise video-EEG recording.29 Few studies have focused on the problem of implementing a “reproducible” methodology to allow TLE surgery with an optimal cost–benefit ratio, both in terms of seizure outcome/quality of life and costs.29 However, even in the best published TLE surgical series, only 80% of patients have “good outcome”, often including patients in Engel class II who are not seizure free.4 Moreover, several non-invasive protocols include very expensive procedures, such as MRI volumetric analysis, functional MRI, single photon emission computed tomography, positron emission tomography, and the Wada test. This reduces the cost effectiveness of surgery for epilepsy and the number of patients to whom it can be offered.28–31

    Another open issue concerns the identification, by means of non-invasive techniques, of different TLE subsyndromes (mesial, lateral, and mesiolateral TLE)32 so that different surgical strategies can be offered as appropriate. In fact, the “classic” surgical approach in TLE tends to consider standard surgical procedures (temporal lobectomy or selective amygdalohippocampectomy),29,33,34 with the risk of removing cortical regions not involved in epileptogenesis or not resecting the whole epileptogenic zone.

    When MRI shows a tumour with a brief (less than two years) seizure history, we consider ictal video-EEG recording unnecessary when the patient and eyewitness history, clinical semiology, and/or interictal EEG are concordant with the tumour site.35 Recurrent seizures are recognised as a potential cause of neuronal damage, and hence have been hypothesised to modify the extent of the epileptogenic zone in time.36 Therefore, in tumorous epilepsy with a brief history, it is more likely that the epileptogenic zone is restricted to the lesion, and surgery may be limited to lesionectomy alone.

    The results of lesionectomy alone for the treatment of TLE as a result of a temporal lobe mass lesion have been controversial. Jooma and colleagues37 and Cascino and colleagues38 reported that lesionectomy led to a “seizure free” condition in only 20% of patients with temporal lobe mass lesions, whereas Fried and colleagues39 reported a seizure free outcome in 80% of patients.35 Kraemer et al reported that 73% of the patients affected by TLE as a result of vascular malformations were seizure free (no seizures in the last year of follow up, according to Dukes’s classification40 after lesionectomy).35 In general, it is agreed that if the seizure disorder exists for more than a year before surgical resection, lesionectomy alone may not be efficacious.37,41 Other studies42,43 corroborate the assumption that a shorter seizure disorder duration correlates with a better prognosis when lesionectomy is the surgical approach.

    In our series, 14 patients with TLE had a short seizure history and MRI evidence of a temporal mass lesion: three of these patients were excluded from “fast” lesionectomy and were admitted to diagnostic level 2 for video-EEG recording because of lesional pathology not colocalised with EEG interictal findings, whereas 11 patients had lesionectomy with excellent outcome (all in Engel class 1a). Two of the three abovementioned patients who were entered into diagnostic level 2 had extensive temporal lobectomy (one in Engel class 1a and one in Engel class 1b), whereas the other patient had lesionectomy (Engel class 1a).

    Although some studies44 report that ictal video-EEG recording may be not mandatory in patients with TLE, we decided to admit most lesional or non-lesional patients with TLE to video-EEG recording. This strategy is obviously “obligatory” in patients with normal MRI or absent or bilateral interictal EEG epileptiform abnormalities. Moreover, in our opinion, a tailored operation cannot be offered without having ictal EEG and clinical data, which allow the three different TLE subsyndromes to be identified. Therefore, we developed a diagnostic grid for TLE surgery based on correlations among anatomical, electrical, and clinical data, in accordance with principles first planned by Bancaud and Talairach,9 and later developed by their successors,10–12 obtained by non-invasive investigations. According to many studies that evaluated the localising value of a single diagnostic criterion,22,23,45–52 the data obtained from presurgical evaluation were grouped in clusters strongly suggesting the location of epileptogenesis in the mesial or lateral aspect of the temporal lobe. Then, based on our experience, specific combinations of different criteria were used for localisation. Although most authors agree about the localising value of clinical signs and of neuroradiological findings in suggesting the epileptogenic zone, to our knowledge, only a few clinical studies53 have used EEG ictal patterns for localising purposes. In our classification scheme, we particularly stressed the localising value of the ictal scalp EEG pattern in TLE, according to the theories of Pacia and Ebersole,23 based on “the most comprehensive systematic study of scalp–intracranial ictal EEG findings in patients with temporal lobe epilepsy”.54 Their findings confirmed and extended previous studies on this issue,55,56 revealing that > 5 Hz ictal EEG pattern over the temporal regions is the characteristic pattern associated with mesial temporal seizures. In contrast, there is comparatively little information about neocortical temporal lobe seizures, and in most cases the available descriptions are shown in negative terms, emphasising the absence of features typical of mesial temporal lobe seizures. In light of this, the studies of Ebersole’s and Foldvary’s groups should still be thought of as important contributions; moreover, they depicted neocortical temporal lobe seizures in positive terms. Both authors agree that low frequency scalp EEG ictal patterns in temporal lobe seizures are highly related to the lateral neocortical seizure onset zone. Compared with other non-invasive diagnostic protocols31,33,57 for TLE surgery, aimed at selecting those patients who should benefit from standard temporal lobectomy and identifying those patients who should be investigated invasively or excluded, our diagnostic grid enables TLE classification to be refined into the mesial, temporal, and mesiotemporal subtypes. Hence, surgery can be specifically planned, as far as possible, according to the extension of the epileptogenic zone in the three different TLE subtypes (that is, AMTL in mesial-TLE, ETL in mesiolateral-TLE, and lesionectomy in lesional lateral-TLE), avoiding unnecessarily extensive surgery. As suggested by studies based on stereotactic intracerebral EEG recordings,58 during which the simultaneous involvement of both the amygdala, the hippocampus, and the temporal pole at the onset of temporal lobe seizures was often observed, the so called “mesial” temporal lobe seizures probably do not always arise solely from the amygdalo–hippocampo–parahippocampal complex. Therefore, in the absence of intracerebral recordings, in mesial-TLE, we decided to use the AMTL resection, which includes the temporal pole in addition to the mesial temporal structures, rather than the selective amygdalo–hippocampectomy.

    The usefulness of our T-Prot was confirmed by the excellent outcome (overall, 85.1% in Engel class 1), although we cannot exclude the possibility that a similar outcome would have been achieved without using a tailored resection. Given that the diagnostic approach we describe is based on a global assessment that correlates and combines a set of criteria, we could not carry out an analysis of the performance of each diagnostic criterion (anatomical, ictal EEG, and ictal clinical) in terms of sensitivity and specificity. In our classification method, it is not possible to assign a patient to a group using single criteria, and this made it impossible for us to test the performance of a single criterion. In any case, the performance of the full set of criteria was very satisfactory.

    Although our findings need to be confirmed by further studies with a longer follow up duration and by randomised controlled trials comparing our protocol with other established protocols, they suggest that in most patients with TLE different subsyndromes can be accurately identified using non-invasive anatomo–electro–clinical data, and can be treated effectively and safely with a tailored operation.


    We thank the nursing staff and the team of neurophysiology technicians (L Cacciola, F Orabona, and R Ponticelli) who performed meticulous video-EEG recordings in all patients.



    • Competing interests: none declared

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.