- Top of page
- Supporting Information
Purpose: There is still controversy in deciding which patients with frontal lobe epilepsy (FLE) should undergo resective surgery, even though it is a well-established therapy. The aim of this study is to define multiple outcome measures and determine whether there are certain subpopulations of preferred surgical candidates that have a more favorable seizure prognosis.
Methods: Fifty-eight patients underwent resective FLE surgery with a mean follow-up period of 79.3 months (range 12–208 months). Patient demographics, clinical seizure characteristics, seizure-onset zone within the frontal lobes, and diagnostic tests were tabulated. Engel class, International League Against Epilepsy (ILAE) class, postoperative seizure patterns, time to first recurrent seizure, and seizures and employment during the last year of follow-up were used as outcome measures. Neuropsychological performance and Beck Depression Inventory (BDI) scores were used to define neuropsychological outcome and examined as predictors of seizure outcome.
Key Findings: Thirty-three (57%) patients with resective surgery had an Engel class I outcome and 29 (50%) had an ILAE class I outcome. Mean time to first seizure after surgery was 33.3 months (range 0–208). Only 14 patients (24%) were completely seizure-free without auras (Engel IA) throughout the entire follow-up period. The most common pattern of seizure recurrence was mixed, with prolonged periods of seizure freedom intermixed with recurrences. In addition, 32% of patients made gains in employment and 52% were able to reduce use of antiepileptic drugs (AEDs), although only 9% discontinued AEDs. No significant association was found between class I or class IA outcome and the presence of a focal magnetic resonance imaging (MRI) abnormality, any specific localization of seizure focus within the frontal lobe, or neuropsychological change.
Significance: Findings indicate that that long-term outcome is generally favorable in FLE resective surgery, and support the need for considering multiple outcome measures to more fully characterize clinically relevant postsurgical changes. Outcome can be favorable even in MRI-negative patients.
Epilepsy surgery is a well-established therapy for frontal lobe epilepsy (FLE). Success rates vary between 20 and 80%, with most studies reporting a favorable seizure outcome of Engel class I or class IA in around 50% of patients (Fish et al., 1993; Salanova et al., 1994; Laskowitz et al., 1995; Zentner et al., 1996a; Kazemi et al., 1997; Ferrier et al., 1999; Janszky et al., 2000; Jobst et al., 2000; Mosewich et al., 2000; Kral et al., 2001; Schramm et al., 2002; Worrell et al., 2002; Jeha et al., 2007; Elsharkawy et al., 2008; Lee et al., 2008). However, outcome measures of those studies are not uniform and often focus on short-term outcome.
It is well established that FLE surgery is less successful than temporal lobe epilepsy surgery with respect to seizure outcome (Yun et al., 2006; Jehi et al., 2010). In patients with FLE, normal magnetic resonance imaging (MRI; Ferrier et al., 1999; Janszky et al., 2000; Jeha et al., 2007; Elsharkawy et al., 2008), generalized scalp electroencephalography (EEG) abnormalities (Janszky et al., 2000), and preoperative seizure frequency (Laskowitz et al., 1995) have been identified as negative predictive factors. Furthermore, resections of the supplementary motor area (SMA) are thought to have a relatively more favorable outcome than resections in other regions within the frontal lobes (MacDougall et al., 2009). Nonetheless, whether resective surgery is worthwhile in patients with FLE, especially those with negative magnetic resonance imaging (MRI) findings, is still a matter of debate (Elsharkawy et al., 2008; Lee et al., 2008). Such surgeries frequently require large intracranial EEG studies, which are a great burden to the patient and not without surgical risk (Nair et al., 2008). As a consequence, various surgical centers have different surgical approaches and indication criteria for FLE surgery.
A potential limitation of many studies investigating outcome following resection for FLE is the reliance on one specific outcome measure, which is commonly freedom from disabling seizures or Engel outcome classification. Such outcome measures are, however, limited in scope with respect to potential clinically and personally meaningful changes resulting from surgery. Other factors that are rarely examined in surgical studies but are of potential salience include such outcomes as significant reduction in seizure frequency, several years of seizure freedom despite later recurrence, gains in employment, driving ability, and side effects of AEDs (Knowlton et al., 2011).
The present study evaluated multiple outcome measures following surgery for FLE including several seizure outcome measures, employment, and reductions in AED. Furthermore, neuropsychological performance and depression were also examined, as they have been reported to affect well-being more significantly than seizure outcome (Metternich et al., 2009).
- Top of page
- Supporting Information
Patients with a diagnosis of intractable FLE and subsequent intracranial EEG study and/or FLE surgery were enrolled into the study. For patients who did not have invasive EEG monitoring, the decision to perform resective surgery was based on scalp EEG monitoring and a corresponding lesion in noneloquent cortex. Postoperative follow-up had to be well-documented for a minimum of 12 months.
Presurgical evaluation at the Dartmouth-Hitchcock Comprehensive Epilepsy Center included comprehensive seizure history, scalp video-EEG monitoring, high resolution MRI, neuropsychological testing, ictal single-photon emission computed tomography (SPECT), positron emission tomography (PET), and intracranial EEG, as deemed necessary in a multidisciplinary conference. Information was also collected on sex, risk factors, seizure focus and lateralization, type of surgery, number of surgeries, patient age at surgery, age at first seizure, duration of epilepsy prior to surgery, and type of preoperative seizures. Seizure foci were categorized as orbitofrontal, frontal convexity, frontopolar, or SMA/medial frontal. Localizations in the SMA and other medial frontal regions were combined for the purpose of analysis because 17 patients had seizures localized specifically to the SMA, whereas only three patients had seizures localized to the other medial frontal regions.
Follow-up data were obtained via repeated clinic visits. Postoperative neuropsychological testing was performed a minimum of 5 months after surgery.
The MRI included high-resolution anatomic scans and fluid-attenuated inversion recovery (FLAIR) imaging, using the best available techniques at our center at the time of evaluation. Most patients were scanned with a 1.5 T MR scanner, whereas a subset seen since 2007 was scanned using a 3.0 T magnet.
Skilled nursing personnel knowledgeable in EEG interpretation performed ictal SPECT studies. Injections were considered valid only if they occurred within 30 s of the first clinical or electrical sign of the seizure. In nonlesional, difficult to localize patients or in patients with questionable injection times, ictal SPECT studies were repeated. Twenty-three patients had more than one ictal SPECT. Ictal SPECT studies were subtracted from interictal SPECT studies and coregistered on MRI (SISCOM). Ictal SPECT results were used to guide electrode placement. Decisions about the extent of resection were not based on ictal SPECT.
Ictal SPECT, interictal scalp EEG, ictal scalp EEG, and PET results were classified as either concordant or not concordant with the area resected. More specifically, the test results were correlated with seizure focus lateralization, lobe localization, and sublocalization within the frontal lobe. Intracranial EEG was examined in the 51 patients with available information. Focal onset was defined as clearly involving ≤6 electrode contacts at the first electrical sign of seizure activity. If >6 electrodes or larger cortical regions were involved, onset was considered regional. EEG patterns were classified as low amplitude, fast activity, which has previously been shown to be associated with a favorable outcome (Wetjen et al., 2009), or as rhythmic, higher amplitude activity including periodic spiking.
Seven patients had more than one resective procedure, with an average time of 26.8 ± 11.2 months between surgeries. For the purposes of this study, follow-up information was tabulated for the period after the most recent resective surgery.
Seizure-related outcome measures were tabulated based on yearly follow-up data. This included Engel class and subclasses (Engel et al., 1993), the International League Against Epilepsy classification system (ILAE; Wieser et al., 2001), time to the patient’s first postoperative seizure, and whether the patient was free of disabling seizures during the last year of follow-up. The latter measure considers only the last 12 months of available follow-up, as compared to ILAE class outcome, which was assigned based upon seizure outcome during the last 12-month period after the surgical anniversary. Engel classification, on the other hand, considers the entire follow-up period. The time to the first postoperative seizure was charted for use in Kaplan-Meier analysis.
Because postoperative seizure frequency fluctuated yearly for some patients, seizure outcome was also assessed through the perspective of the postoperative seizure pattern. Seizure pattern categories included free of disabling seizures, running-down, running-up, mixed, and seizure recurrence. The free of disabling seizure pattern was assigned to patients who never had another disabling seizure (auras were allowed), whereas those assigned the recurrent seizure pattern experienced seizure relapse starting less than 1 year after surgery. The running-down pattern applied to patients who experienced seizure recurrence for over 1 year initially after surgery, before seizures began decreasing in frequency until the patient was free of disabling seizures and remained so for at least 1 year at the end of follow-up. The opposite running-up pattern involved initial seizure freedom for at least 1 year before seizures increased in frequency for at least 1 year through the end of follow-up. Lastly, the mixed seizure pattern involved alternating periods of seizure freedom and seizure relapse, with each period lasting at least 1 year. All outcome data were collected by personnel (SL) not involved in clinical care of the patients.
Postoperative employment was compared to preoperative employment. Part-time work, defined as 20 h/week, and full-time work were treated equally as employment. Being a full-time student was considered equivalent to being employed.
Neuropsychological test data were available for 39 patients. Because this study is retrospective in nature, the tests were not given at a standard time before or after surgery. All tests, however, occurred within 2 years before or after surgery. Testing occurred an average 307.5 ± 176.8 days after surgery (n = 26). Furthermore, patients did not all complete the same measures due to variations in referral issues at the time of testing, patient specific issues (e.g., fatigue), and changes in test versions (e.g., CVLT to CVLT-II) over the 19-year period of data collection.
Intellectual functioning was assessed with the Wechsler Adult Intelligence Scale (WAIS; Wechsler, 1981). Memory was examined with the California Verbal Learning Test Total Trials 1–5 and Long-Delay Free Recall (CVLT; Delis et al., 1987, 2000), and the Logical Memory subtest from the Wechsler Memory Scale (WMS; Wechsler, 1987, 1997). Language was examined with the Boston Naming Test (Kaplan et al., 1983) and phonemic fluency as determined using either the Controlled Oral Word Association Test (Spreen & Benton, 1977) or the Verbal Fluency subtest from the Delis-Kaplan Executive Function System (DKEFS; Delis et al., 2001). Processing speed was assessed using Trail Making Test Part A (Reitan & Wolfson, 1985) or Condition 2 from DKEFS Trail Making (Delis et al., 2001). Executive functions were assessed with the Trail Making Test Part B (Reitan & Wolfson, 1985) or Condition 4 from DKEFS Trail Making (Delis et al., 2001), as well as perseverative errors on the Wisconsin Card Sorting Task (Heaton, 1981). Preoperative (n = 32) and postoperative (n = 18) scores from all of the preceding tests were converted into z-scores to facilitate amalgamation across test versions and comparisons between measures.
Depression was assessed using the Beck Depression Inventory (BDI; Beck, 1987; Beck et al., 1996). BDI scores were analyzed in a slightly different subpopulation (preoperative n = 39; postoperative n = 23) in order to maximize the number of patients available for analysis.
Postsurgical reduction in number of antiepileptic drugs (AEDs) or reduction in AED dosage was also considered as an outcome measure. The number and dosage of AEDs at the last follow-up was compared to the number and dosage of AEDs taken immediately before surgery.
Univariate and multivariate statistical analyses were utilized to test for predictors of outcome. Exploratory univariate statistical analysis was performed first and involved the use of chi-square tests, Fisher’s exact test when small group size precluded the use of chi-square analysis, and paired t-tests. Kruskal-Wallis tests were also utilized for analysis of depression scores. Multiple logistic regression was then utilized to test for significant predictors of seizure, employment, and neuropsychological outcome. Kaplan-Meier analysis was used to calculate the probability of remaining Engel class I throughout the follow-up.
The study was approved by the Dartmouth College Institutional Review Board/Committee for the Protection of Human Subjects.
- Top of page
- Supporting Information
Of the 71 FLE patients, 63 (89%) underwent intracranial EEG monitoring prior to surgery. Four (6%) had resections without intracranial EEG. This reflects the referral bias to the epilepsy center. A resectable seizure focus was identified in 58 patients (82%) (Fig. 1). Of the 58 patients who underwent resection, 32 (55%) were lesional and 26 (45%) were nonlesional.
Of the 13 patients (18%) who did not have a resectable seizure focus, 5 underwent no surgery, 5 underwent callosotomies, and 3 underwent multiple subpial transection (MST).
Analysis from this point forward refers to the population of 58 resective patients.
Demographic, clinical, and diagnostic information is presented in Table 1. Seizure localizations are included in Table 2. The mean follow-up period was 79.3 months (range 12–208).
Table 1. Demographic, clinical, and diagnostic information
| ||N = 58a (%)|
|Seizures lateralized to left hemisphere||32 (55%)|
|Mean age at surgery (years) ± SD (range)||29.8 ± 11.8 (9–58)|
|Mean age at first seizure (years) ± SD (range); (n = 56)||10.1 ± 8.6 (0–40)|
|Mean duration of epilepsy (years) ± SD (range); (n = 56)||19.1 ± 10.9 (1–49)|
| Seizures in family, first degree||6 (10%)|
| Seizures in family, second degree||7 (12%)|
| Perinatal complications||5 (9%)|
| Febrile seizures||4 (7%)|
| Meningitis/encephalitis||1 (2%)|
| Head trauma with loss of consciousness (LOC)||6 (10%)|
| Head trauma without LOC||12 (21%)|
| Sensory||12 (21%)|
| Motor/tonic||38 (66%)|
| Generalized tonic–clonic||26 (45%)|
|History of status epilepticus (convulsive or nonconvulsive)||19 (33%)|
|MRI nonlesional||26 (45%)|
| Normal||18 (31%)|
| Gliosis/encephalomalacia||7 (12%)|
| Tumor||6 (10%)|
| Vascular malformation||4 (7%)|
| Dysplasia||23 (40%)|
|PET scan performed||12 (21%)|
| Abnormal PET||3 (25%)|
| PET abnormality concordant with lateralization of resection||2 (67%)|
| PET abnormality concordant with localization of resection||1 (33%)|
| PET abnormality concordant with sub-localization of resection||1 (33%)|
|Ictal SPECT performed||44 (76%)|
|Mean number of ictal SPECTs per patient ± SD (range)||1.6 ± 1.6 (0–6)|
|First ictal SPECT concordant with area resected (n = 44)|
| Lateralization||38 (86%)|
| Localization to lobe||30 (68%)|
| Localization to sublobe||22 (50%)|
|Subsequent ictal SPECT(s) concordant with area resected (n = 23)|
| Lateralization||22 (96%)|
| Localization to lobe||18 (78%)|
| Localization to sublobe||13 (57%)|
|Any ictal SPECT concordant with area resected (n = 44)|
| Lateralization||41 (93%)|
| Localization to lobe||35 (80%)|
| Localization to sublobe||28 (64%)|
|Ictal SPECTs mutually concordant with area resected (n = 23)|
| Lateralization||19 (83%)|
| Localization to lobe||14 (61%)|
| Localization to sublobe||8 (35%)|
|Scalp interictal EEG concordant with area resected (n = 57)c|
| Lateralization of epileptiform or slowing abnormalities||24 (42%)|
| Localization of abnormalities to frontal lobe||30 (53%)|
|Scalp ictal EEG concordant with area resected (n = 53)c|
| Lateralization of epileptiform or slowing abnormalities||22 (42%)|
| Localization of abnormalities to lobe||34 (64%)|
| Mean number of seizures recorded (n = 54)||11.5 ± 10.0 (0–53)|
|Intracranial EEG monitoring||54 (93%)|
| Single intracranial EEG monitoring session||41 (71%)|
| Several intracranial EEG monitoring sessions||13 (22%)|
| Mean number of electrodes (n = 54) ± SD (range)||86.3 ± 32.7 (16–169)|
| Mean number of arraysd (n = 54) ± SD (range)||7.0 ± 3.5 (2–19)|
| Mean number of grids (n = 54) ± SD (range)||2.2 ± 1.4 (0–5)|
| Bilateral electrode placement (n = 54)||26 (48%)|
| Interhemispheric grids or strips implanted (n = 54)||29 (54%)|
| Mean number of intracranial seizures recorded(n = 51) ± SD (range)||13.2 ± 14.9 (0–87)|
| Focal onset (n = 51)||32 (62.7%)|
| Regional onset (n = 51)||19 (37.3%)|
| Low amplitude onset (n = 51)||20 (39.2%)|
| Rhythmic onset (n = 51)||31 (60.8%)|
| Both a focal and low amplitude onset (n = 51)||14 (27.5%)|
Table 2. Assignment of Engel outcome classes and another seizure outcome measure, seizure freedom during the last year of follow-up, to the 58 resective patients
|Location||Total patient #||Class I||Class II||Class III||Class IV||Class IA||Seizure-free last 12 months of follow-up|
|Orbitofrontal||13 (23%a)||9 (69%b)||3 (23%)||1 (8%)||0 (0%)||4 (31%)||8 (62%)|
|SMA + medial frontal||20 (34%)||11 (55%)||4 (20%)||3 (15%)||2 (10%)||6 (30%)||12 (60%)|
|Frontal convexity||15 (26%)||8 (54%)||2 (13%)||3 (20%)||2 (13%)||1 (7%)||7 (47%)|
|Frontopolar||10 (17%)||5 (50%)||0 (0%)||3 (30%)||2 (20%)||3 (30%)||5 (50%)|
|All resective patients||58 (100%)||33 (57%)||9 (16%)||10 (17%)||6 (10%)||14 (24%)||32 (55%)|
Assigned Engel outcome classes are presented in Table 2. Thirty-three (57%) patients had a class I outcome, which corresponds to freedom from disabling seizures for at least 2 years.
Only 14 patients (24%) had an Engel class IA outcome, which equates to no event ever after surgery and does not allow for subjective auras.
Other seizure outcome measures
Twenty-nine patients (50%) were assigned an ILAE class 1 outcome after being completely seizure-free without auras during the last 12 months following the surgical anniversary. Other ILAE outcome assignments are presented in Table 3.
Table 3. Assignment of ILAE outcome classes to the 58 resective patients
|ILAE class||Outcome descriptiona||Number of patients|
|1||Seizure-free, no auras||29 (50%)|
|2||Only auras, no other seizures||4 (7%)|
|3||1–3 seizure days per year, allowing for auras||6 (10%)|
|4||4 seizure days per year to 50% reduction in number of seizure days, relative to preoperative baseline||11 (19%)|
|5||<50% reduction of seizure days to 100% increase in seizure days, relative to preoperative baseline||5 (12%)|
|6||>100% increase of seizure days, relative to preoperative baseline||6 (2%)|
Thirty-two patients (55%) were free of disabling seizures, allowing for auras, during their last year of follow-up.
The mean time to the first postoperative seizure was 33.3 months (range 0–208). A Kaplan-Meier survival analysis for all patients is displayed in Fig. 2 (left). The majority of relapses occurred during the first 36 months after surgery. Patients who had achieved a class I outcome at 100 months (>8 years) after surgery were likely to remain class I for the rest of the follow-up period.
Figure 2. (A) This Kaplan-Meier survival curve presents the probability that patients will maintain a class I outcome over the follow-up period. (B) These Kaplan-Meier survival curves present the probability that patients, either MRI normal (blue) or abnormal (green), will maintain a class I outcome over the follow-up period.
Download figure to PowerPoint
Seizure sublocalization failed to correlate significantly with class IA or class I outcomes (p > 0.3; p > 0.7; Table S1).
The presence of a lesion on MRI did not correlate with Engel outcome, suggesting that patients who were MRI negative were not less likely to be seizure-free or to have a class I outcome than MRI-positive patients (p = 0.287; p = 0.672). Seventeen (53%) of 32 lesional patients versus 15 of 26 (58%) nonlesional patients were free of disabling seizures during the last year of follow-up. According to Kaplan-Meier survival analysis (Fig. 2, right), the survival curves for MRI normal versus abnormal patients maintaining a class I outcome throughout the follow-up period overlap, corroborating with the lack of a significant difference in outcome.
Twenty-eight (88%) of the 32 lesional patients underwent complete lesion resection. The other four patients who did not had more than one lesion prior to surgery, and underwent resection of only one lesion. One of the four was completely seizure-free, whereas the others had Engel class II (n = 2) and III (n = 1) outcomes.
There was not a statistically significant difference between patients with normal pathology versus abnormal pathology in terms of achieving either seizure freedom (Engel IA; p = 0.513; Table S1) or a class I Engel outcome (p = 0.199; Table S1). Specific types of pathology (Table 1) similarly did not associate with Engel outcome, although sample sizes were small. Fifteen (65%) of 23 patients with dysplasia achieved class I outcomes. Of the 15 patients (26%) that were MRI and pathology negative, five (33%) achieved a class I outcome, whereas two (13%) of those patients were completely seizure-free (IA).
Variables related to intracranial monitoring such as the type of implant and number of electrodes did not correlate with seizure outcome (Table S1). Similarly, intracranial seizure-onset patterns (Table 1) were not associated with Engel class outcome. Preoperative ictal SPECT findings of abnormalities in the frontal lobe were related to Engel class outcome if multiple ictal SPECTs were performed (p = 0.046; Table S1).
Interictal EEG findings were not significantly associated with Engel class outcome. Some ictal EEG findings were relevant to outcome. If the ictal scalp EEG results were concordant with the lateralization of the seizure focus, the patients were more likely than others to have a class IA outcome (p = 0.026; n = 53). Twenty-two patients (42%) had an ipsilateral scalp EEG ictal onset, whereas 15 (28%) had bilateral onset, 5 (9%) had contralateral onset, and 11 (21%) exhibited no ictal abnormalities. Other scalp EEG findings did not reach statistical significance.
The only risk factor to correlate with outcome was head trauma (p = 0.044), and patients with this risk factor were less likely to have a class IA outcome. Other prognostic factors did not reach significance with regard to a class IA or class I outcome (Table S1).
Details about the univariate analysis of prognostic factors are presented in Table S1.
The most common postoperative seizure pattern was free of disabling seizures (found in 33% of patients), which is different from an Engel class IA outcome because it allows for auras. Four patients (7%) experienced the running-down pattern, whereas ten (17%) experienced the running-up pattern. Thirteen (22%) experienced a mixed pattern and 12 (21%) were assigned the recurrent seizure pattern.
Preoperative and postoperative employment information was available for only 56 of the 58 resective patients. Two patients were excluded from this analysis due to missing postoperative employment information.
Thirty-four (61%) of the 56 patients were employed preoperatively. A total of 35 patients (63%) were employed at last follow-up. Of the 34 patients employed preoperatively, 28 (82%) were employed at the last follow-up. Four of the six patients who were considered employed prior to surgery but who were no longer considered employed after surgery were actually students prior to surgery. One student with seizure recurrence (class IIIA) could not work. The other three students either returned to classes or started working after surgery, yet they stopped taking classes or working once seizures recurred. The other two patients who lost employment at last follow-up were able to work during extended postoperative periods of seizure freedom, but experienced seizure recurrence before the last follow-up and lost employment. Therefore, employment loss is related to seizure recurrence. Of the 22 patients (39%) who were unemployed prior to surgery, seven (32%) gained employment at the last follow-up.
Postoperative employment was associated with a class I outcome (p = 0.01). Therefore, patients who did not have class I outcomes were more likely to not be employed following surgery.
Based on repeated measures statistical analysis, patients assigned a Engel class I outcome compared to those assigned class II–IV outcomes were not more likely to be less depressed or have a higher full scale IQ after surgery compared to before surgery (p > 0.05). There were also no significant differences between the groups on the other neuropsychological measures presurgically, and there was a significant Group X Time interaction irrespective of how seizure outcome was classified (i.e., Engel class I versus all other classes). We performed a Kruskal-Wallis analysis based on Beck Depression Index (BDI) scores and the seizure-onset localization within the frontal lobe. There was no significant relationship between preoperative or postoperative depression score and seizure-onset zone, if examined independently. However, there was a significant change between preoperative and postoperative depression score stratified by seizure-onset localization (p = 0.031). Based on post hoc analysis, the SMA/medial frontal and frontopolar patients showed an increase in BDI score postoperatively, whereas the orbitofrontal and frontal convexity patients showed a decrease in BDI score postoperatively.
Preoperative AED information was available for all 58 patients, but postoperative AED information was available for only 57 patients. Five (9%) of the 57 patients were off all AEDs at last follow-up. The mean number of AEDs taken immediately prior to surgery was 2.17 (n = 58) and the mean number of AEDs taken at the last follow-up was 1.89 (n = 57).
Sixteen patients (28%) took a reduced number of AEDs at last follow-up, whereas 9 (16%) took an increased number of AEDs. Fifteen patients (26%) remained on identical medications as preoperatively, with eight (14%) on reduced doses of those same AEDs.
Vagus nerve stimulation
Eight (14%) patients underwent vagus nerve stimulation (VNS) after resective surgery. VNS was placed on average 48 months after the resective surgery.
Three of the patients (38%) who had postoperative VNS had some seizure frequency improvement. Three (38%) had no worthwhile improvement, while two (25%) experienced an increase in seizure frequency after the VNS implantation.
Three resective patients (5.2%) experienced complications and were readmitted. Two had cerebrospinal fluid (CSF) leaks and one had an infection. All three complications resolved completely, and two of those patients had a class I outcome. After SMA resection, 12 of the 17 SMA-localized patients (71%) experienced hemineglect on the contralateral side, which is expected after SMA resection and not considered a complication (Zentner et al., 1996b). All resolved completely and patients were preoperatively informed about this risk.
- Top of page
- Supporting Information
Our observation of an overall Engel class I outcome of 57% and a combined Engel class I and II outcome of 73% is consistent with previous reported studies (Fish et al., 1993; Salanova et al., 1994; Laskowitz et al., 1995; Zentner et al., 1996a; Kazemi et al., 1997; Ferrier et al., 1999; Janszky et al., 2000; Mosewich et al., 2000; Kral et al., 2001; Schramm et al., 2002; Worrell et al., 2002). Although 50% of patients achieved a class 1 ILAE outcome, which equates to complete seizure freedom during the last year of follow-up, only 24% of patients were completely seizure-free without auras during the entire follow-up (Engel class IA). AEDs were discontinued in only 9% of patients. This demonstrates the difficulties of defining meaningful outcome after epilepsy surgery. Is aggressive epilepsy surgery only justified if the probability of a complete cure is high or is some palliation of devastating seizures enough of an indication? Every detail of outcome measures is relevant for meaningful comparison of multiple studies.
In our series, patients with negative MRI did have outcomes equally good as patients with definite lesions, which contradicts most previous reports (Cascino et al., 1992; Lorenzo et al., 1995; Zentner et al., 1996a; Smith et al., 1997; Ferrier et al., 1999; Mosewich et al., 2000; Kral et al., 2001; Jeha et al., 2007; Elsharkawy et al., 2008; Lee et al., 2008). We could not find a definite predictive factor for a good outcome, despite the extensive analysis examining preoperative studies, clinical characteristics, pathology, and extent of resection. The only predictive preoperative finding was concordance between lateralized scalp ictal EEG and outcome. There are a few other studies that include nonlesional patients and also report good outcomes in that population (O’Brien et al., 2000; Knowlton et al., 2008). Those investigators rely heavily on functional studies such as ictal SPECT. O’Brien et al. (2000) found that resection of the SPECT focus (SISCOM) was predictive of outcome, independent of MRI findings. Ictal SPECT, magnetic source imaging, and PET were found to be predictive of seizure-free outcome in nonlesional patients in another study (Knowlton et al., 2008). We could not find a definite association of ictal SPECT findings with favorable outcome, in contrast to those studies. This may be an effect of sample size. Because ictal SPECT was utilized mainly for electrode planning, ictal SPECT certainly guided intracranial investigation.
Seizure recurrence patterns after surgery were mixed in our study. Jehi et al. (2010) attempted to clarify if one postoperative seizure amounts to seizure recurrence. The likelihood of recurrence increased with the number of postoperative seizures (Jehi et al., 2010). These investigators reported a running-down effect (several postoperative seizures before seizure freedom) in 11%, which is similar to 7% in this study and others (Elsharkawy et al., 2008; Jehi et al., 2010). The running-down pattern was the least common observed postoperative seizure pattern in this study, suggesting that the running down of seizures is possible but unlikely for patients who face initial seizure recurrence. There is increasing discussion about late seizure relapses after the initial 2-year postoperative period (Schwartz et al., 2006). In this study, the majority of seizure relapses occurred in the first 3 years after surgery, and the likelihood of remaining seizure-free after a prolonged period of good outcome is high.
A mixture of prolonged periods of seizure freedom with periods of seizure recurrences was the most frequent postoperative pattern in our study. Although this pattern is certainly influenced by postoperative AED changes, it seems obvious the AED changes alone would not have had the same effect if not combined with resective surgery. The frequent occurrence of postoperative mixed seizure patterns in our study questions the usefulness of measures that chart the time to first seizure recurrence (Burneo et al., 2008). Previous studies have shown that only complete seizure freedom significantly improved quality of life in patients with epilepsy (Birbeck et al., 2002). But, aren’t patients also likely to significantly benefit if there are prolonged periods of seizure freedom, despite some seizure relapse?
Aside from seizure freedom, gains in employment and reduction of AED side effects may have a similar if not greater impact on the patients’ lives. Unfortunately, standardized quality of life measures were not universally available for most of our patients, but epilepsy surgery resulted in some employment gains and reflects previous findings that employment gains correlate with better seizure outcome (Chin et al., 2007). In our study, most losses of employment at the last follow-up occurred in patients who were initially employed after surgery but then lost employment once seizures recurred. Employment is a difficult measure of outcome, because it also depends on a variety of psychosocial factors that we could not systematically analyze, such as the availability of employment opportunities and the economy. The number of AEDs or the dosage of AEDs in patients with unchanged medications could be reduced in 42% of patients. This likely resulted in a beneficial reduction of side effects, though this was not tested systematically. The retrospective nature and uncontrolled design of this study make it difficult to interpret those results, but reduction in AEDs has been shown to positively influence patients’ quality of life (Elsharkawy et al., 2009). Decisions to effectively reduce or change AEDs after epilepsy surgery still remain a matter of debate. Over time it has become less common to completely discontinue AEDs for fear of late recurrences (Schiller et al., 2000), which is reflected in our study, given that only a few seizure-free patients discontinued medications.
None of the prognostic variables tested are significant predictors of outcome. The location of epilepsy surgery within the frontal lobe has no influence on outcome, refuting our clinical impression that SMA resections are associated with a better outcome than resections in other parts of the frontal lobes. Previous findings that generalized EEG discharges predict seizure outcome could not be confirmed (Janszky et al., 2000). In various outcome studies, prognostic factors are inconsistent (Schramm et al., 2002; Jeha et al., 2007; Elsharkawy et al., 2008), suggesting that more research is required to determine consistent predictors of outcome.
Neuropsychological and neuropsychiatric consequences of epilepsy surgery are of importance, but are less well-described for FLE than for TLE surgery. In the present investigation, preoperative neuropsychological scores did not predict seizure outcome, and surgery was not associated with significant change in neuropsychological functioning. Other investigations have yielded inconsistent changes, with some reporting memory improvement in seizure-free patients (Helmstaedter et al., 1998; Lendt et al., 2002), although at least mildly worse executive functions related to surgery have also been noted (Helmstaedter et al., 1998; Morris & Cowey, 1999). The heterogeneity of findings may be partly due to differences in the specific frontal regions that were resected (Helmstaedter et al., 1998; Risse, 2006).
There was no correlation between depression scores and seizure outcomes in this study. A previous study including both TLE and FLE patients explored the relationship between resective surgery and depression, concluding that lower preoperative BDI scores were correlated with better seizure outcomes (Metternich et al., 2009). These results were challenged by a later study that found no significant correlation between preoperative BDI and seizure outcome in epilepsy surgery patients (Hoppe et al., 2010). When we analyzed depression scores with respect to seizure-onset zone, there was no correlation preoperatively or postoperatively, if examined independently. When we analyzed changes in preoperative and postoperative depression scores, the patients with mesial frontal or frontopolar resection experienced a postoperative worsening of their depression. This is an interesting finding, as mesial frontal structures have been implicated to mediate depression and are now the target for treatment of depression with brain stimulation (Holtzheimer et al., 2012). The subgroups in our study were small, so definite conclusions cannot be made. We conclude that preoperative depression should not exclude patients from surgery, but the complex relationship between preoperative and postoperative depression in epilepsy surgery warrants further standardized investigation (Witt et al., 2008; Hamid et al., 2011; Wrench et al., 2011; Jobst, 2012).
This study certainly has multiple limitations. It is a retrospective review. Small sample sizes in subgroup analysis may mask subtle effects. In addition, the retrospective analysis of neuropsychological outcome was inherently limited because not all patients took the same batteries and updated versions of certain tests were released during the retrospective period.
The study is meant to provide some guidelines to counsel patients who consider FLE surgery. It certainly reflects common clinical practice. In conclusion, after considering all variables, FLE surgery in all locations within the frontal lobe improves seizure outcome, but psychosocial outcome and comorbidities such as depression need more investigation. Epilepsy surgery should be offered to FLE patients, even if MRI is negative. Overall outcome can also be favorable even if the patient is not completely seizure-free.