Abstract
Background: Because of the relatively poor results of frontal lobe epilepsy (FLE) surgery, identification of prognostic factors for surgical outcome is of great importance.
Methods: To identify predictive factors for FLE surgery, we analyzed the data of 61 patients (mean age at surgery 19.2) who had undergone presurgical evaluation and resective surgery in the frontal lobe. Postoperative follow-up ranged from 0.5 to 5 years (mean 1.78). Fifty-nine patients had MRI-detectable lesions. Histopathologic examination showed dysplasia (57.4%), tumor (16.4%), or other lesions (26.2%). Thirty postoperatively seizure-free patients were compared with 31 non–seizure-free patients with respect to clinical history, seizure semiology, EEG and neuroimaging data, resected area, and postoperative data including histopathology.
Results: Three preoperative and two postoperative variables were related to poor outcome: generalized epileptiform discharges, generalized slowing, use of intracranial electrodes, incomplete resection detected by MRI, and postoperative epileptiform discharges. The only preoperative factor associated with seizure-free outcome was the absence of generalized EEG signs. Multivariate analysis showed that only the absence of generalized EEG signs predicts the outcome independently. Moreover, the occurrence of a somatosensory aura, secondarily generalized seizures, and negative MRI was identified as additional independent risk factors for poor surgical results.
Conclusions: The absence of generalized EEG signs is the most predictive variable for a seizure-free outcome in FLE surgery. Furthermore, nonlesional MRI, somatosensory aura, and secondarily generalized seizures are risk factors for poor surgical results.
Frontal lobe epilepsy (FLE) is the second largest group of localization related epilepsies occurring in 6 to 30% of all surgically treated epilepsy patients.1,2 Compared with temporal lobe epilepsy (TLE), the surgical outcome for FLE is less favorable.2-4 This may be because the lateralization and localization of the epileptogenic zone in the frontal lobes are apparently less successful with the available methods used in the presurgical workup.2,4-8 Moreover, eloquent areas, especially in the dominant frontal lobe, prohibit a complete resection even when they have been found to be epileptogenic. Nevertheless, owing to the introduction of long-term video-EEG, subdural electrodes, and modern neuroimaging techniques, the localization and excision of the epileptogenic zone have become possible in more and more drug-resistant patients with FLE, and the surgical results have become more promising.2,9,10
Because of the less favorable outcome results of FLE surgery, the determination of prognostic factors is important in the prediction of seizure outcome and in decisions regarding the use of intracranial electrodes or the extension of surgery. In addition, the identification of prognostic factors may improve general understanding of the pathophysiology of postoperative failure and of spatial extension of the epileptogenic zone.
Multivariate analysis techniques are favorable for identifying variables that predict surgical outcome. Because TLE is the most frequent form of focal epilepsy with a good surgical prognosis, studies investigating the surgical outcome in focal epilepsy include mostly TLE patients,11-13 whereas FLE patients make up only a small portion of these patients, encompassing 9 to 24%.
Only one study investigated the prognostic factors of FLE surgery in 37 patients by using univariate and multivariate methods. However, neither MRI nor long-term video-EEG were performed on most of the patients included in the study.9 This is a general problem of multivariate studies considering surgical outcome in focal epilepsy.11-13 During the last years it has become evident that the best prognostic factor for epilepsy surgery (including FLE) is a lesion detected by MRI.14-18 Predictive factors found by these studies should be reevaluated to determine whether they are independent of MRI results. For TLE surgery, this question has been recently investigated.19 However, it is not known in FLE which preoperative factors can be used to predict the outcome independently of the MRI-detected lesions and whether long-term video-EEG results provide prognostic information. Hence, our study was aimed at identifying predictive factors for the outcome of FLE surgery observing data of 61 patients who underwent high-standard presurgical evaluation including MRI and long-term video-EEG.
Methods.
Presurgical evaluation at the Epilepsy Surgery Department of the Epilepsy Center Bethel.
Because observed variables were based on presurgical evaluation of epileptic patients, our protocol is briefly described below.
In patients who were considered possible candidates for epilepsy surgery, a detailed neurologic history was obtained. Therapy resistance to first-line antiepileptic drugs was evaluated. A high-resolution MRI was performed as a rule. In patients with negative MRI, a second and more detailed MRI investigation was performed on the basis of other results of presurgical evaluation. Patients underwent continuous video-EEG monitoring lasting 3 to 10 days. In all patients except for small children, psychiatric and neuropsychological examination, as well as a social assessment, was performed. Findings of presurgical evaluation were discussed at a multidisciplinary case-conference, where decisions were made concerning the possibility and type of surgery, the possible epileptologic, psychosocial, and neuropsychological outcome, as well as the necessity for further investigations with intracranial EEG recordings.
Patients who had undergone epilepsy surgery were reexamined 6 months and 2 and 5 years later with an assessment of epileptologic, psychiatric, neuropsychological, and social outcome. Postoperatively, at least one routine EEG and one MRI were performed.
Patient selection.
For this study we included all patients who had undergone presurgical evaluation at our center, had a resective surgical procedure involving the frontal lobe between January 1992 and December 1998, and returned for at least the 6-month follow-up examination. The patients had undergone lesionectomy, extended lesionectomy, topectomy, subtotal lobectomy, or frontal lobectomy. Patients with nonresective neurosurgery (corpus callosotomy, multiple subpial transsection), with hemispherectomy, and with a history of previous epilepsy surgery were excluded. Sixty-one patients met inclusion criteria. All the patients had undergone a detailed clinical history, MRI, and long-term video-EEG with ictal and interictal recordings. Forty-eight patients had undergone intracranial EEG monitoring with subdural electrode arrays. Clinical and demographic data are provided in table 1.
General data
Data collection.
We investigated the following groups of variables: general data, seizure semiology, EEG and neuroimaging data, localization of resected area, underlying pathology, and postoperative data. General data were obtained from the clinical history and medical records. The list of variables are presented in table 1. The variables of seizure semiology were based on video-EEG recordings, whereas aura, the presence of secondarily generalized tonic-clonic seizures (SGTCS), number of seizure types, and seizure frequency were obtained from the clinical history (see list of variables in table 2). EEG and MRI data (table 3) were determined from noninvasive continuous video-EEG monitoring, preoperative and postoperative high-resolution MRI, and postoperative routine EEG. Preoperatively 32-64 channel EEG recordings were used; electrodes were placed according to the 10-10 system20; the number of electrodes and their placement varied individually corresponding to the suspected epileptogenic region and side. The location and frequency of interictal epileptiform discharges (IED) were assessed by visual analysis of interictal EEG samples of 2 minutes duration per hour. Except in two cases a postoperative EEG was performed 6 to 24 months after surgery using 21 channels, placed according to the 10-20 system including FT9 and FT10 electrodes of the 10-10 system. We categorized the frontal spikes ipsilateral, according to whether they were concordant with the resection side, or contralateral, if they were not. We used the term generalized slowing when any pathologic generalized theta/delta activity was present on the EEG. For statistical purposes, we did not divide patients with generalized pathologic slow waves into more specific subgroups, such as continuous or intermittent slow activity, irregular or rhythmic slow waves (e.g., intermittent rhythmic delta activity), due to the small number of cases in our statistical evaluation. Patients were classified into having no generalized signs if neither generalized slowing nor generalized IED was present on the EEG. In most patients, preoperative MRI was made by a Siemens Magnetom Impact 1.0-T scanner and included T1-weighted three-dimensional volume, proton density, and T2-weighted images. In 49 cases, a postoperative MRI was performed with the same protocol. The localization and lateralization of resection are presented in table 4 and in the figure. Underlying pathology (table 5) was verified by histologic examination of resected tissue. Focal cortical dysplasia was further divided according to the presence of balloon cells considered to be indicating a higher degree of pathology. Owing to the small number of cases, tumors were not divided into more specific groups. Table 6 shows the results of postoperative MRI and EEG.
Seizure semiology
EEG and MRI data
Localization of resected area
Figure. Schematic description of the four frontal areas. Patients were divided into four groups according to the localization of resected area (see table 4).
Pathologic findings
Postoperative findings
Selection of variables.
The above-mentioned variables were chosen for their presumed importance in presurgical evaluation or if they were previously reported as a predictive factor of epilepsy surgery. The results of psychiatric, neuropsychological, and sociological assessments were not included owing to the heterogeneity of the population in terms of age (26% of patients were below age 10). Intraoperative electrocorticography data were ignored as well, although their prognostic value had already been verified21 since most of the patients had undergone preoperative invasive monitoring.
Outcome assessment.
Outcome was assessed at regular postoperative visits. We only evaluated the seizure outcome, whereas psychiatric, social, and neuropsychological outcome were ignored. For this study, we divided the patients into the two categories, seizure (and aura) free or non–seizure free.
Statistical methods.
For the univariate analysis of the categorical data, χ2, Fisher’s Exact and binomial tests were performed. For comparisons of continuous variables the Mann-Whitney U test was used. To identify which variables could predict the outcome independently, we performed a stepwise logistic regression for variables demonstrating a significant effect in univariate analyses. To evaluate the individual predictability of outcome we computed a discriminant analysis including all categorical variables presented in tables 2 to 5⇑⇑⇑ (altogether 31 variables). Use of intracranial electrodes was not included because it was considered to be a nonbiologic variable based on subjective decisions and is therefore not a reliable predictive factor. Postoperative variables (see table 6) were not included in the multivariate analyses. Two-tailed error probabilities smaller than p < 0.05 were considered to be significant. In univariate analysis, we did not alter this limit owing to the large number of variables because the multivariate analysis was assumed to be the final statistical tool in identifying the independent predictive factors.
Results.
Thirty of 61 patients (49%) became seizure free after the FLE surgery. Five patients (8%) had rare disabling seizures postoperatively and were categorized into the non–seizure-free group. The mean follow-up time was 1.78 years (range 6 months to 5 years). Nine patients had a follow-up time <1 year: 3 seizure free and 6 non–seizure free. The follow-up time did not differ between both groups (see table 1). The gender distribution showed a male predominance (34.4% women versus 65.6% men, p = 0.02). This difference had no influence on outcome.
Univariate analysis.
The following preoperative variables showed a significant association with the outcome (see tables 1 to 5⇑⇑⇑⇑): generalized IED, generalized slowing, no generalized signs on the EEG, and use of invasive electrodes. The absence of generalized signs was associated with a seizure-free outcome, other variables with a non–seizure-free outcome.
Two postoperative variables were associated with poor surgical results: the incomplete resection according to postoperative MRI and postoperative IED (table 6). In eight patients the postoperative MRI showed an incomplete resection. In two cases, the incomplete resection was because of the large extension of pathology. In the other six patients the extension of the pathology was not recognizable intraoperatively, and the incomplete resection was only recognized by the postoperative MRI. All of the latter six patients had a focal cortical dysplasia. Five patients with incomplete resections underwent a reoperation, two of them became postoperatively seizure free.
Independence of variables and multivariate analysis.
A forward stepwise logistic regression was performed investigating three presurgical factors that were found significant by univariate analysis. Only the absence of generalized EEG signs correlated independently with surgical outcome. The presence of generalized slowing or generalized IED did not. This was not surprising since the first variable p includes the latter two ones. Moreover, there was also a correlation between generalized slowing and generalized IED (p = 0.0006, Fisher’s Exact Test).
The presence of postoperative IED also showed an association with MRI-detected incomplete resection of pathology (p = 0.03, Fisher’s Exact Test). According to discriminant analysis, four variables can significantly predict the seizure outcome: 1) no generalized signs on the EEG, 2) somatosensory aura, 3) SGTCS and 4) MRI-detected pathology. Somatosensory aura and secondarily generalized seizures were independently associated with non–seizure-free outcome, the other two with seizure-free outcome. Although sensory aura, SGTCS, and MRI-detected pathology showed no significant correlation with outcome by univariate analysis, their addition to the model significantly improved the prediction of outcome. Discriminant analysis resulted in the following equation: 
−0.55×SGTCS−0.48×SAURA
Where NGS = 1 if no generalized signs are present on the EEG, and NGS = 0 if generalized IED or slowing is present,
MRPAT = 1, if MRI shows pathology; 0 if not
SGTCS = 1, if the patient has SGTCS; 0 if not
SAURA = 1, if somatosensory aura is present; 0 if not.
When (PO) is above 0, then a seizure-free outcome is predicted. Negative values predict a non–seizure-free outcome.
By using this equation it was possible to correctly classify 79% of the 61 patients: 90% of the seizure-free patients and 68% of the non–seizure-free patients. Conversely, if a patient within our investigated population had a PO > 0, then he had a 73% chance of seizure freedom. When PO < 0, then this probability was only 13.5%.
Discussion.
In our study, four preoperative variables were associated with postoperative outcome: generalized IED, generalized slowing, no generalized signs on the EEG, and use of invasive electrodes. The association between use of chronic invasive EEG recordings and poor outcome in focal epilepsy surgery is well-known12 and reflects either incongruent data from noninvasive procedures or the need for determination of eloquent frontal cortex such as rolandic or Broca’s areas. Therefore, we did not consider this variable as a predictive factor. Regarding the other three preoperative variables, we found that only the absence of generalized EEG signs (IED or slowing) could be held as an independent predictive factor.
For individual prediction of surgical outcome, a discriminant analysis resulted in an equation consisting of four variables. Using this equation (see Results), we could correctly predict the outcome in 79% of the cases. This equation contains four variables: absence of generalized EEG signs, presence of MRI pathology, somatosensory aura, and the presence of SGTCS. The first two factors correlated with seizure-free outcome, the second two with poor outcome. Except for the absence of generalized EEG signs, the other three variables were not found to be significant by univariate analysis. However, their inclusion in this multivariate model improved the prognosis significantly. This could be expected because the variable absence of generalized signs explains most of the variance.
Because all but two patients had MRI-detectable lesions (96.7%), which were also confirmed by neuropathologic investigations, there was no correlation between MRI findings and the outcome according to univariate analysis. This ratio is higher than the proportion of cases with lesional MRI reported in other studies.2,4,17 This reflects the fact that by using high-resolution MRI, morphologic abnormalities could be detected more often. Especially focal cortical dysplasia diagnosed in 57% of patients can only be detected by high-resolution MRI.22,23 Nevertheless, a selection bias has to be suspected. In the past years it has become evident that an MRI-detected lesion is the best prognostic factor for epilepsy surgery. Therefore, we preferred lesional cases, and patients with negative MRI scans were rarely selected for surgery. Because almost all of our patients had a lesional MRI, our study could identify variables that could predict the surgical outcome independent of lesional MRIs, but not of negative MRIs.
Ipsilateral and contralateral frontal, extrafrontal, multifocal, and generalized epileptiform activity can appear in FLE.2,5,6,24 Presence of generalized spikes was found to be a poor predictive factor in two multivariate studies dealing exclusively or mainly with TLE patients.13,25 Other studies that pertain to epilepsy surgery in general,11,26 hemispherectomies,27 or FLE surgery,9 found no association between the outcome and generalized IED. Our results showed that the presence of generalized IED correlated with poor surgical outcome but not independently of generalized slowing. These two variables were closely related.
Generalized IED in focal epilepsy occurring most frequently in FLE1,6 is termed secondary bilateral synchrony28 (SBS). The pathomechanism of SBS is not clear, but it is likely that discharges from the primary focus spread through the thalamus or through the corpus callosum to the corresponding contralateral area, producing a nearly synchronized discharge.28-30 An alternative explanation is that SBS arises from an interaction of multiple active foci.31 In focal epilepsy, de novo appearance of SBS with drug-resistant tonic axial seizures suggests a progressive nature of epileptogenesis.32 Our findings of a correlation between generalized epileptiform activity with poor outcome support the assumption that SBS represents a more widespread epileptogenic area. This may be a result of secondary epileptogenesis or may be owing to a generalized epileptic feature also present to various extent in focal epilepsies.33
Our results showed that generalized slowing correlated highly with generalized spikes and poor outcome, but according to the multivariate analysis, it was not independent of the absence of generalized EEG signs in predicting outcome.
In univariate analysis we found no correlation between seizure semiology and surgical outcome. According to the discriminant analysis, somatosensory aura and SGTCS significantly improved the prediction as negative prognostic factors. Somatosensory aura is the most common type of aura in FLE.6 Nevertheless, its presence might refer to an extrafrontal (postcentral) symptomatogenic area. In focal epilepsy surgery it was associated with poor operative results.26 In two studies, there was an association between SGTCS and poor surgical outcome,26,34 whereas other studies did not find such a correlation.9,11-13
In our patients, both postoperative IED and incomplete resection detected by postoperative MRI correlated with poor outcome. However, they could not be held as prognostic factors since they were obtained parallel to postoperative seizure evaluation. This association has already been reported in focal epilepsy surgery.35 Most of the patients with incomplete resections showed by postoperative MRI had focal cortical dysplasia, in which it is often difficult to delineate the exact extension of this abnormality preoperatively and intraoperatively.
We considered 6 months as the lower limit for follow-up time. This limit may seem to be too early to assess the surgical outcome, and most studies deal with a longer (1 to 2 years) minimal follow-up time. Nevertheless, some other studies have evaluated the surgical outcome at 6 months.15 According to a systematic observation of the seizure outcome over time, the outcome assessment after 6 months is an excellent indicator of long-term surgical results.36 Moreover, in extratemporal epilepsy the long-term relapse rate is much lower compared with the rate in TLE.37,38
The duration of epilepsy, age at onset, the localization of the resected area, and histopathologic findings showed no influence on the surgical outcome. The orbito-polar localization is associated with a favorable outcome (see table 4), but this was not significant, probably because of the relatively small number of orbito-polar cases. This trend may reflect that this region is relatively far from the eloquent areas, therefore more extensive resections can be performed without risk of postoperative neurologic deficit.
Only one recent study aimed at identifying independent prognostic factors in FLE surgery. Observing data of 37 patients, Ferrier at al. found that focal neuroimaging abnormality independently correlated with good outcome, whereas contralateral headversion was associated with poor surgical results.9 In their study, there was no correlation between generalized IED and outcome. Generalized slowing was not investigated. One explanation for the difference between our results and those of Ferrier and coworkers may be that they included patients with rare disabling seizures in the good-outcome group. Thus, in their good-outcome group, only 60% of the patients were completely seizure free. The other explanation could be that most of their patients did not undergo a comprehensive presurgical evaluation: Less than 50% of their patients had MRIs, and only 40% had long-term video-EEGs. Some of them had no EEG data at all. In accordance with our data, they showed that the duration and onset of epilepsy, age at surgery, and the pathologic findings did not correlate with surgical outcome.
Acknowledgments
Supported in part by a grant from the Deutsche Forschungsgemeinschaft (DFG, Eb 111/2-2) and from the Society for Epilepsy Research Bethel (J.J.).
Acknowledgment
The authors thank the members of the epilepsy surgery program Bethel, especially Dr. H. Pannek (neurosurgery), Dr. J. Tuxhorn (pediatric epileptology), and Prof. R. Lahl (neuropathology).
- Received September 7, 1999.
- Accepted December 17, 1999.
References
Disputes & Debates: Rapid online correspondence
REQUIREMENTS
If you are uploading a letter concerning an article:
You must have updated your disclosures within six months: http://submit.neurology.org
Your co-authors must send a completed Publishing Agreement Form to Neurology Staff (not necessary for the lead/corresponding author as the form below will suffice) before you upload your comment.
If you are responding to a comment that was written about an article you originally authored:
You (and co-authors) do not need to fill out forms or check disclosures as author forms are still valid
and apply to letter.
Submission specifications:
- Submissions must be < 200 words with < 5 references. Reference 1 must be the article on which you are commenting.
- Submissions should not have more than 5 authors. (Exception: original author replies can include all original authors of the article)
- Submit only on articles published within 6 months of issue date.
- Do not be redundant. Read any comments already posted on the article prior to submission.
- Submitted comments are subject to editing and editor review prior to posting.