Summary: Purpose: To review the clinical, neurophysiologic features and surgical outcomes in patients with frontal lobe tumors and chronic intractable seizures.
Methods: Medical records of patients with intractable epilepsy who underwent resection or stereotactic biopsy of frontal lobe tumor (confirmed by surgical pathology) seen between 1985 and 1999 at Yale University School of Medicine Epilepsy Center were reviewed for age at diagnosis, age at onset of seizures, delay between seizure onset and tumor diagnosis, types and frequencies of seizures, EEG results, use of anticonvulsants, extent of surgery, pathological diagnosis, and tumor recurrence.
Results: Thirty-seven patients were included. Mean age at seizure onset was 31.6 years, and at tumor diagnosis was 36.2 years. Mean duration between onset of seizures and tumor diagnosis was 6.1 years. Seventeen patients had auras. Seizure frequency averaged 7.6 seizures per week, with 58% of patients having more than one seizure type. All patients used anticonvulsants, with 90% eventually using polytherapy. All patients eventually underwent at least one surgical procedure. Only 13 (35.1%) patients were class I. Twelve (32.4%) patients were class II, seven (18.9%) class III, and five (13.5%) class IV. No statistically significant differences were seen between good and poor long-term seizure outcome in relation to specific tumor pathology, seizure types, or type of resection.
Conclusions: Long-term surgical outcomes in tumoral frontal lobe epilepsy are more favorable than those in nontumoral intractable frontal lobe epilepsy (65% class I or II) and less favorable than those in other tumoral epilepsy (overall, 70% class I). Frontal location of intracranial neoplasm may predict a less favorable long-term epilepsy prognosis than tumoral epilepsy in general, an observation for which several explanations are proposed.
Intracranial neoplasms are found in 3–15% of patients with chronic epilepsy (1–3). The epilepsy in patients with brain tumors is challenging with regard to both the understanding of the epileptogenic mechanisms triggered by the tumor and the criteria of surgical management (4,5). The goal of surgery in these patients is the resection of the tumor as well as the cure of epilepsy (4–6).
The reported incidence of seizure reduction in intractable epilepsy patients with tumors after surgery varies between 36.4 and 92%(4,7,8), but most reported patients had temporal lobe tumors (4,5,8–11). Frontal lobe epilepsy accounts for 15 to 30% of intractable focal epilepsies (12). Reported incidence of seizure freedom or significant seizure improvement after surgery in frontal lobe epilepsy varies between 23 and 80%(13–17). Patients with lesional frontal lobe epilepsy, however, have better surgical outcomes than those with frontal lobe epilepsy without lesions (15–18). Even studies of lesional frontal lobe epilepsy treated with surgery had few patients with neoplasms (15–23). We report a retrospective study on the long-term seizure outcome in patients with frontal lobe tumoral epilepsy treated with surgery.
Medical records of all patients with medically intractable partial epilepsy who underwent resection or stereotactic biopsy of frontal lobe tumors (confirmed by surgical pathology) seen between January 1985 and December 1999 at Yale University School of Medicine Epilepsy Center were reviewed. Data collected included age at tumor diagnosis, age at onset of seizures, delay between seizure onset and tumor diagnosis, types and frequencies of seizures, EEG results, use of anticonvulsants (AEDs), extent of surgery, pathological diagnosis, and tumor recurrence. Follow-up was ≥1 year for inclusion.
All patients had preoperative high-resolution brain magnetic resonance imaging (MRI), and only one patient required long-term intracranial monitoring for localization to proceed with surgical treatment .In the others, the structural lesion on MRI was locationally consistent with other findings supporting ictal onset from the area of the tumor including clinical seizure semiology, interictal EEG localization, focal or regional seizure onset recorded with surface ictal EEG, and neurocognitive deficits consistent with frontal lobe damage.
To achieve seizure control, gross total tumor resection was attempted, guided by intraoperative confirmation of tumor-free margins with frozen section analysis. Postoperative imaging and pathological examination of permanent tissue margins were also used to establish extent of resection. A subtotal resection, or biopsy, was performed when resection posed an unacceptable risk of morbidity, based on intraoperative or extraoperative functional mapping. Electrocorticography (ECoG) was performed, but did not change the extent of resection.
Factors in relation to long-term surgical outcomes including tumor location, pathology, seizure types, extent of surgery, and EEG findings were examined statistically by using Fisher's Exact test . Two-tailed t test and analysis of variance (ANOVA) were used to evaluate the association between age at seizure onset and age at tumor resection with seizure outcome.
Seizure outcome was graded according to Engel's classification (24) as follows: class I, seizure free (or seizures only in the setting of AED modification or noncompliance); class II, rare seizures; class III, meaningful improvement in seizure frequency or severity; and class IV, no meaningful improvement in seizure frequency or severity. AEDs were tapered in seizure-free patients only after ≥12 months of seizure freedom, with the plan to resume if seizure recurred (maintaining class I outcome).
Of the 20 female and 17 male patients identified, epilepsy began at a mean age of 31.6 years (1–74 years; Table 1). Mean age at tumor diagnosis was 36.2 years (4–76 years), and mean duration between onset of seizures and tumor diagnosis was 6.1 years (2 weeks to 36 years). Seizure occurrence prompted the evaluation leading to tumor diagnosis in 32 (86.5%) patients, with six patients seen after a first generalized tonic–clonic convulsion.
Table 1. Patient clinical data
|Age at seizure onset (yr)||31.6 (1–74)|
|Age at tumor diagnosis (yr)||36.2 (4–76)|
|Time between seizure and tumor diagnosis||6.1 (2 wk to 36 yr)|
|Follow-up duration (yr)||8.7 (1–15 yr)|
|Aura||17 patients (45.9%)|
|One seizure type||16 patients (43.2%)|
|simple partial seizures||24 patients (64.9%)|
|Complex partial seizures||13 patients (35.1%)|
|Generalized tonic–clonic seizures||15 patients (40.5%)|
|Average total seizure frequency||7.6 seizures/week (four per year to 15 per day).|
Seventeen (45.9%) patients exhibited auras. The most common was a rising epigastric sensation in four (23.5%) of 17 patients. Fear and light-headedness were the aura in two (11.8%) patients each. Other auras were described as gasping throat sensation, funny feeling all over, neck pain, heavy sensation on the arm, not feeling well, sensory aura all over the spine, and frontal head pain. Twenty-two (59.5%) patients had more than one type of seizure. Simple partial seizures were the most common seizure type and were seen in 24 patients. Generalized tonic–clonic seizures were seen in 15 patients; 10 of those had fewer than three generalized seizures. In four of the 15 patients with generalized seizures, they were exclusively nocturnal. Complex partial seizures were seen in 13 patients and were the only seizure type in two patients. Psychogenic seizures, in addition to simple partial motor seizures and complex partial seizures, were diagnosed in one patient. Average seizure frequency was 7.6 seizures per week (one seizure every 4 months to 15 seizures a day). Status epilepticus was seen in four patients. Two of those patients had epilepsia partialis continua as the first sign of the tumor.
In 22 (59.5 %) patients, the tumor was in the left frontal lobe, whereas in 15 (40.5%) patients, it was on the right. Tumor locations within the frontal lobe were: dorsolateral, 13 patients; supplementary motor (SMA), eight patients; frontopolar, five patients; motor cortex, four patients; cingulate, three patients; and orbitofrontal, two patient. In two patients, the tumor involved an extensive region of the frontal lobe.
Oligodendroglioma was the most common tumor, diagnosed in 12 (32.4%) patients, followed in frequency by low-grade glioma (grade I or II) in 10 (27.0%) patients. Eight (21.6%) patients had meningiomas, three of which were atypical. Dysembryoplastic neuroepithelial tumor (DNET) was found in three patients. The others had anaplastic astrocytoma, ganglioglioma, mixed astrocytoma–oligodendroglioma, and intracranial schwannoma (one patient each).
Surface EEG recordings were available for review in 29 (78.4%) of 37 patients. Of the 29 EEGs reviewed, four (13.8%) were normal. Interictal frontal epileptiform discharges were seen in three (10.3%) patients, and frontal slowing on the side of the tumor in seven (24.1%) patients. One patient had a seizure during surface EEG with focal rhythmic discharge over the ipsilateral frontal lobe. Bifrontal, falsely localizing, or generalized findings were seen in eight (27.6%) recordings. Abnormal findings in these 25 EEGs are summarized in Table 2.
Table 2. Surface EEG findings in 25 frontal tumor patients
Continuous prolonged closed-circuit video-EEG monitoring with surface electrodes was performed in our Epilepsy Monitoring Unit in 18 (49.6%) patients. Results of monitoring are summarized in Table 3. Interictal epileptiform discharges (spikes or sharps) were seen in eight (44.4%) patients. Frontal and frontocentral epileptiform discharges were seen in four (22.2%) patients. Fifteen (83.3 %) patients had at least two ictal events during the monitoring. Of the 15 with recorded seizures, frontal ictal onset was seen in two (13.3%) patients, whereas congruent frontotemporal or frontocentral ictal onset was seen in four (26.7%) patients. Congruent lateralizing ictal onset was seen in two (13.3%) patients. Combining interictal and ictal findings, 10 (55.6%) patients had additional localizing and lateralizing information from video-EEG monitoring. In the remaining patients, clinical seizure semiology and interictal EEG findings were consistent with the tumor location and were thought to provide sufficient evidence for surgery without EEG monitoring.
Table 3. Prolonged EEG monitoring results
|1||L||CPS||Dorsolateral||Bifrontal synchronous spikes and polyspike discharge||L. frontotemporal buildup|
|2||L||SPS||Dorsolateral||L. frontotemporal slowing||No EEG changes|
|3||R||SPS||SMA||R. frontal slowing||No seizures|
|4||L||SPS||SMA||Normal||Diffuse left hemispheric theta buildup|
|5||L||GS||Frontopolar||L. anterior temporal sharps||Poorly localized L. hemispheric changes|
|6||R||CPS||Cingulate||R. frontal sharps||Several seizures, no focal onset|
|7||L||SPS, CPS||Frontopolar||L. frontal slowing||No seizures|
|8||R||GS||Dorsolateral||Normal||R. frontal rhythmic beta buildup with rapid secondary generalization|
|9||L||CPS, GS||Dorsolateral||L-sided sharp waves||Few CPS, EEG nonlocalizing|
|10||L||SPS||Motor strip||Normal||No EEG changes|
|11||L||CPS||Dorsolateral||Bifrontal independent spikes more in the L. frontal regions||Few seizures, nonlocalizing|
|12||R||SPS||Motor strip||Rare delta slowing over rt hemisphere||Few seizures, no EEG changes|
|13||L||CPS||Motor strip||Left frontal slowing||No seizures|
|14||R||CPS||Frontopolar||R. frontal spike and wave||Focal motor seizures R. frontal semirhythmic delta|
|15||L||CPS||Motor strip||Left frontocentral spikes||L. frontocentral buildup|
|16||R||CPS||SMA||Normal||Nonlocalizing with bifrontal onset|
|17||L||CPS||Dorsolateral||Left frontocentral spikes||Five seizures, three with L frontocentral onset|
|18||L||CPS||Frontopolar||Midtemporal sharps||Rt fronto- and midtemporal region onset|
All tumors were demonstrated by high-resolution MRI. Gadolinium enhancement was seen in 10 patients. Two patients had previous normal computed tomography (CT) studies. Positron emission tomography (PET) scanning was done in five patients: two were normal, two had frontal hypometabolism in the tumor area, and one had hypometabolism in the ipsilateral frontotemporal area. Interictal single-photon emission computed tomography (SPECT) scan was done in four patients: one was normal, two had focal frontal changes, and one had congruent frontoparietal decrease in blood flow.
All patients eventually underwent at least one surgical procedure. Twenty-three patients had gross total resection of the tumor. Four patients had gross total resection followed by radiotherapy, three patients had subtotal resection, and five patients had subtotal resection followed by radiotherapy. One patient underwent stereotactic biopsy followed by radiotherapy, and one patient was treated with radiotherapy before being seen in our institution, and then with subtotal resection and chemotherapy.
In an average follow-up of 8.7 years, nine (24.3 %) patients had recurrence of tumor or malignant transformation of a lower-grade tumor after surgery. Three patients with atypical meningioma had recurrence after initial gross total resection. Recurrence was treated with resection and radiotherapy. Six patients with low-grade gliomas (three astrocytomas and three oligodendrogliomas) had malignant transformation after initial resection. Four of those patients had worsening seizures as the presenting symptom of malignant transformation. All patients were treated with partial resection and radiotherapy or chemotherapy.
Before surgery, patients used one to six AEDs (average, 2.8 drugs/patient). Each patient received AED monotherapy initially, but ultimately only four (10.8%) patients were maintained on their initial AED. Before surgical intervention, 11 (29.7%) patients had their seizure frequency improved by ≥50% with the use of AEDs. Of the AEDs used, 23% had to be discontinued because of side effects. Table 4 summarizes reported experience with AEDs used in eight or more patients.
Table 4. Anticonvulsant use, efficacy, and side effects
|Phenytoin||32||11 (34.4%)||7 (21.9%)|
|Carbamazepine||33||12 (36.4%)||8 (24.2%)|
|Phenobarbital||23||11 (47.8%)||4 (17.4%)|
|Valproic acid||18||3 (16.7%)||5 (27.8%)|
|Gabapentin||14||3 (21.4%)||2 (14.3%)|
|Lamotrigine||10||2 (20%)||2 (20%)|
|Topiramate||8||1 (12.5%)||2 (25%)|
|Others||16||4 (25%)||5 (31.3%)|
Long-term seizure outcome and predictors
Average follow-up after surgical intervention was 8.7 years (range, 1–15 years). Thirteen (35.1%) patients were seizure free or had only auras (class I); 12 patients were class II (32.4%); seven (18.9%) were class III; and five (13.5%) patients had no improvement. Patients with excellent or good outcome (classes I and II) were compared with patients with no significant improvement classes (III and IV), Table 5. No statistically significant differences between the two groups were seen for any of the listed predictive factors or diagnostic result.
Table 5. Analysis of seizure outcome postoperatively comparing classes (I and II) with classes (III and IV)
|Number of patients||25||12||—|
|Age at seizure onset (mean) (yr)||35.3||26.5||0.37|
|Age at tumor diagnosis (mean) (yr)||36.8||34.7||0.34|
|Tumor hemisphere (right/left)||9/16||6/6||0.49|
|Tumor location: dorsolateral (13)||7||6||0.27|
|Tumor location: SMA (8)a||7||1||0.23|
|Low-grade glioma (11 patients)||5||6||0.12|
|Oligodendroglioma (12 patients)||9||3||0.71|
|Meningioma (8 patients)b||7||1||0.23|
|Total resection with or without RTX (26)||19||7||0.44|
|Subtotal resection with or without RTX (10)||6||4||0.69|
|Frontal or regional interictal epileptiform changes: present/absent (33 patients)||8/14||2/9||0.43|
|Long-term EEG (18 patients): done/not done||10/15||8/4||0.24|
Patients with no seizure improvement
Of the 12 patients with no significant seizure improvement (classes III and IV),two patients refused further evaluation. Four patients had seizure worsening with malignant transformation of lower grade tumors and were treated with reoperation and radiotherapy or chemotherapy with some improvement. Six patients with no tumor recurrence were reevaluated for persistent seizures. In four of those, seizures originated from the vicinity of the tumor, whereas in other two, seizure onset was distant from the resected vicinity. One patient had reoperation with minor seizure improvement; one patient underwent vagal nerve–stimulator implantation with some seizure improvement; the remaining four patients refused further intervention.
Tumoral frontal lobe epilepsy accounts for almost a third of refractory frontal lobe epilepsy (25). In this retrospective, nonrandomized study, we report our experience in 37 patients with pathologically verified tumoral frontal lobe epilepsy, who were surgically treated and had >1 year of follow-up. Seizures of frontal lobe origin are generally brief, frequent, and often occur in clusters (26). Almost half the patients in our group had auras, a figure somewhat similar to prior studies (19,20). Mean age of seizure onset was 31.6 years, clearly representing an older population than patients with nontumoral frontal lobe seizures (13,19,23), but closer to the age of patients with tumoral epilepsy in general (3,5,27). Mean duration of seizures before tumor diagnosis was 6.1 years, substantially less than mean duration mentioned in prior studies (3,4,6). This probably reflects the earlier manifestations of seizures with motor content and hence earlier diagnosis in patients with tumors in the frontal lobe compared with other locations.
All patients were diagnosed with MRI scan. Functional imaging was done only in a small number of patients (PET and SPECT scan). PET studies have not been well studied in extratemporal epilepsy. The reported percentage of regional hypometabolism in patients with frontal lobe epilepsy varies from 45 to 60%(22,28).
EEG studies in the diagnosis of frontal lobe epilepsy have limitations because of the anatomic size, extensive networks, and limited access to scalp EEG recording of the frontal cortex (25). In our patient population, interictal frontal epileptiform discharges were seen in 10.3%, and congruent focal frontal slowing, in 24.1% of the patients. More than a fourth of the patients had bifrontal, falsely localizing, or generalized findings. In previously published studies, interictal EEG had poor sensitivity and specificity for localization, with focal or localizing value in 13–72% of the patients (19,29–31). Quesney (28) reported that bifrontal or generalized bilaterally synchronous epileptiform potentials were the most prevalent interictal pattern in 38% of 148 patients (28).
Prolonged video-EEG monitoring may be underused in patients with tumoral epilepsy. Patients with tumoral epilepsy often have epilepsy surgery without ictal recordings (45,46). In our group, 50% of the patients did not undergo prolonged video-EEG monitoring before resective surgery. Patients who did not have monitoring were those patients with less frequent seizures, concordantly localizing seizure semiology, and localizing surface EEG. Among the patients studied, prolonged EEG monitoring showed focal or lateralizing ictal onset in slightly more than half the patients who had seizures recorded, and additional localizing or lateralizing information was obtained by monitoring in 55.6% of the patients. Previous reports regarding localizing information with prolonged EEG monitoring in frontal lobe epilepsy were highly variable, with yields ranging from 4 to100%(12,18,19,30–32). According to the literature, more obvious ictal EEG features can distinguish seizures originating in dorsolateral frontal versus medial frontal lobe, an observation originally made on some of these patients (29).
The prognostic significance of ECoG in patients with frontal lobe tumoral epilepsy has not been specifically addressed in prior studies (42). In a series reported by Montreal Neurological Institute, patients with nontumoral frontal lobe epilepsy had favorable long-term seizure outcome if epileptiform discharges were localized on preoperative ECoG (42). Patients with tumoral epilepsy (temporal and extratemporal) reported from our center showed a trend for an association between postresection spikes on ECoG and seizure recurrence (43). As shown in multiple studies, however, complete structural resection better predicted seizure control than did complete or incomplete spike focus resection (44). In this series, resection was determined by tumor-free margins whenever possible. Although ECoG was performed, and tailoring resection toward spike foci was not pursued, ECoG findings did not suggest an alternative approach.
Few studies addressed AED use in patients with tumoral epilepsy (5,9,10). Retrospective analysis of the use of AEDs has multiple limitations (33). Generally limited efficacy and an excess of side effects were observed in patients with medically refractory tumoral epilepsy (5,9,10,29). Polytherapy was eventually used in 90% of our patients. Improvement in seizure frequency was noted more with the use of traditional AEDs [phenytoin (PHT), carbamazepine (CBZ), and phenobarbital (PB)]; however, this probably reflects the fact that those medications were used as first- or second-line therapy, whereas other AEDs were used in more resistant cases. Almost a fourth of patients had to discontinue their AEDs because of side effects, an incidence higher than typically reported with nontumoral epilepsy (33).
Favorable surgical outcomes for frontal lobe epilepsy varied in reported studies from 23 to 80%(13–23), with larger studies reporting 50–55% favorable outcomes. (16,17,34–36). Although abnormalities on neuroimaging predicted better surgical outcomes in more than one study (11–17,37,38), in general, few series found reliable predictors of surgical outcome in patients with frontal lobe epilepsy (15–22). Most patients included in prior published series of epilepsy surgery in frontal lobe epilepsy had nontumoral frontal lobe epilepsy (15–22). In our patient population, surgical outcome was clearly better than surgical outcome in the whole group of intractable frontal lobe epilepsy patients and much better than in the group of nonlesional frontal epilepsies (13–15); however, surgical outcome was less favorable than that in patients with tumoral epilepsy in other locations (4,7,8), where most recent series reported overall 65–90% favorable results.
One possible explanation for this difference is that most previously reported tumoral epilepsy was in temporal lobe patients (4,7,8). Limitation of resection because of involvement of functional cortex is another possible explanation. Inherent differences in tumor biologic activity and distribution might also explain these differences in surgical results between frontal and otherwise located tumoral epilepsy. The tumor types we found were roughly similar to those in other locations, with the exception that DNET and ganglioglioma represent a greater proportion of temporal lobe tumors associated with intractable epilepsy (11,39,40). Finally, previous authors mentioned the concept of a large epileptogenic zone in frontal versus other lobes (28,41); these results might support this concept of pathophysiologic differences, although the exact nature of such a phenomenon and the nature of epileptogenesis in brain tumors in general remain speculative.