The Spectrum of Long-term Epilepsy–associated Tumors: Long-term Seizure and Tumor Outcome and Neurosurgical Aspects

Authors


  • Accepted February 8, 2003.

Address correspondence and reprint requests to Dr. C. Luyken at Department of Neurosurgery, University of Düsseldorf, Moorenstr. 5, D-40225 Düsseldorf, Germany. E-mail: cordelia.Luyken@uni-dusseldorf.de

Present address of Ingmar Blümcke: Institute of Neuropathology, University of Erlangen, Krankenhausstr. 8-10, D-91054 Erlangen, Germany.

Abstract

Summary: Purpose: To describe the histologic spectrum and clinical characteristics of patients with neuroepithelial tumors and drug-resistant epilepsy and to analyze clinical data and treatment related to seizure outcome and survival.

Methods: Data were analyzed from 207 consecutive patients with intractable epilepsy (aged 2–54 years), who between 1988 and 1999 had ≥50% resection of supratentorial, neuroepithelial tumors. Extent of resection was assessed on postoperative magnetic resonance imaging (MRI); seizure outcome was classified according to Engel's outcome scale; and follow-up data were prospectively updated.

Results: Median follow-up was eight years (range, 2–14 years). Histologic examination revealed 154 classic epilepsy-associated tumors (ganglioglioma, dysembryoplastic neuroepithelial tumor, pleomorphic xanthoastrocytoma, and pilocytic astrocytomas) and 53 others (astrocytomas and oligodendrogliomas). Four World Health Organization (WHO) grade III tumors were found (astrocytoma, n = 3; ganglioglioma, n = 1). After surgery, 82% of the patients were seizure free (class I). The following factors were associated with improved seizure outcome: Short duration of epilepsy before surgery, single EEG focus, absence of additional hippocampal sclerosis or cortical dysplasia, transsylvian approach, other than astrocytomas, and complete tumor resection. After 5 years, only nine (4%) patients had tumor recurrence, four (2%) with malignant transformation and death. None of the four patients with anaplastic tumors died. Even patients with astrocytomas of WHO grade II or III showed 10-year recurrence of only 25% and 10-year survival of 90%.

Conclusions: Tumors associated with long-term epilepsy should be removed early for two different reasons: high rate of seizure freedom and rare but potential risk of malignant tumor progression. The unexpected long survival of these astrocytomas should be investigated by using immunohistochemistry and molecular biology.

In ≤30% of patients with long-term drug-resistant epilepsy, neuroepithelial neoplasms can be encountered (1–5). Histopathologic classification distinguishes two different groups in this cohort. The first contains typical epilepsy-associated tumors such as gangliogliomas (GGs), dysembryoplastic neuroepithelial tumors (DNTs), pleomorphic xanthoastrocytomas (pXAs), and supratentorial pilocytic astrocytomas, WHO grade I, with usually benign behavior (4,6–12). The second group consists of diffuse astrocytomas (Astro) WHO grade II and oligodendrogliomas (Oligo) WHO grade II with 5-year survival rates of 50–65% and a few anaplastic cases classified as WHO grade III with a median survival of 2–3 years (13–24).

The rate of seizure cure in these tumors reaches to ≤82%, and various factors like age at first seizure or age at operation, tumor location, and residual tumor, in former studies only measured by computed tomography (CT) scans, have been reported (5,25–30). Long-term data showing seizure and tumor outcome in these epilepsy-associated tumors are rare with mostly small numbers of cases, and complete tumor resection can be reliably assessed only by postoperative magnetic resonance imaging (MRI).

PATIENTS AND METHODS

In the epilepsy surgery program of the University of Bonn including 841 resective cases between 1988 and 1999, 229 patients were treated for epilepsy and neuroepithelial tumors. These patients with both long-standing pharmacoresistant epilepsy and a tumor seem to constitute a specific group (i.e., long-term epilepsy–associated tumors, LEATs) concerning seizure outcome and influence factors but also concerning the postoperative tumor course.

The aim of this study was (a) to show the different histologic findings in 207 consecutive patients with LEATs, (b) to analyze the rate of seizure cure and factors of influence, and (c) to point out the distinct biologic behavior, especially in the group classified as Astros WHO grade II or III.

Patients and follow-up

All patients from the epilepsy surgery series with neuroepithelial tumors operated on between 1988 and 1999 were extracted from the neuropathological tissue registry and the epilepsy surgery database. Further inclusion criteria were intractable epilepsy for a ≥2 years, supratentorial and hemispheric tumor location, presurgical evaluation during workup in the Department of Epileptology (31,32), resective procedure for both the tumor (≥50%) and the focus, and postoperative MRI to evaluate the extent of tumor removal and the resection of the temporomesial structures.

A series of 214 consecutive patients was accrued. Seven patients had to be excluded, three because of unrelated death, and four because of unavailable follow-up. From the 207 remaining patients, complete and current follow-up was available, usually in the outpatient department, or by information obtained from patients' general practitioners, neurologists, or by phone interview.

The median follow-up was 8 years (2–14 years), and in 180 (88%) of the 203 living patients, a minimum follow-up of 5 years with a median of 8 years was reached.

Surgical techniques

If the epileptogenic focus was incongruent to the lesion, or the hypothesis about the ictogenic area was not clear enough, invasive procedures (e.g., strip, grid, and intrahippocampal depth electrodes) were used, as described previously (30,32–34).

The operative procedures aimed at removing both the epileptogenic focus and the lesion. Resection types included anterior two-thirds temporal lobe resection, transsylvian hippocampectomy plus lesionectomy, temporodorsal hippocampectomy plus lesionectomy, transcortical lesionectomy plus hippocampectomy, and anterior partial (mostly mesial) temporal lobectomy, as described in former reports (30).

Histopathology

Since 1992, histologic diagnoses have been graded according to the WHO 1993 grading system. In the mid-1990s, all cases from before 1992 were reviewed and classified according to WHO 1993 (35–37). Histologic specimens were retrieved from the archive and microscopically reviewed by using hematoxylin and eosin staining as well as immunohistochemical reactions for glial fibrillary acidic protein (DAKO, Glostrup, Denmark), and proliferation antigen Ki67 (Dianova, MIB1, Hamburg, Germany). Paraffin sections (4 μm) were used according to standardized protocols with the avidin-biotin labeling method and diaminobenzidine as chromogen.

Major histopathologic entities included pilocytic or diffuse or anaplastic Astros, Oligos, DNTs, and GGs, the latter being characterized either by a specific glioneuronal element or by a prominent dysplastic neuronal component, respectively, according to the WHO classification for tumors of the nervous system (35).

Seizure outcome

Seizure outcome was classified 1 year after operation and on annual follow-up according to Engel's outcome scale (31). Year-to-year analysis was performed to analyze the rate of seizure recurrence or latter improvement.

Data management

For the purpose of the study, a relational database was built up extracting data from the epilepsy surgery database, the neuropathological tissue registry, and the epileptologic database, mostly comprising prospectively collected data. The following data were included and analyzed for possible influence on seizure and tumor outcome: Age at first seizure, drug resistance, duration of epilepsy, seizure type, invasive evaluation procedures, single or additional EEG focus, location of the tumor, date and mode of resection with or without the hippocampal structures, histology according to WHO 1993, residual tumor evaluated on postoperative MRI, annual seizure outcome, date of seizure and tumor recurrence, recent follow-up data for seizure and tumor outcome, and date and cause of death.

Statistics

The influence of single (categorized) factors on seizure outcome was analyzed by using the χ2 test or Fisher's exact test, if necessary because of the sample size. The method of Kaplan and Meier was used to calculate survival and recurrence rates (38). The comparison of quantitative variables (duration of epilepsy, age at first seizure and at operation) between groups of patients was based on the Wilcoxon–Mann–Whitney test. Stepwise logistic regression was used to select a multifactorial model for seizure outcome from the relevant (p < 0.05) factors found in the one-factorial analysis. Finally we checked whether the explanation of seizure outcome could be improved by the inclusion of terms for interaction between factors. All analyzes were supported by SAS/STAT Software.

Study end points

The end points were set on death or last examination, which in all patients was updated at least until the year 2000. These were with regard to epilepsy: Seizure freedom, persisting seizures, and recurrence and with regard to tumor: Recurrence and death.

RESULTS

Seizure outcome and influence factors

Clinical data and seizure outcome of all 207 patients are demonstrated in Table 1. All 207 patients (male, 104; female, 103) except one with additional homonymous quadrantanopia had drug-resistant epilepsy for ≥2 years (median, 12 years; maximum, 54 years) as the only symptom of the tumor. They had their first seizure at the median age of 13 years (3 months to 62 years) and were operated on at the median age of 28 years (5–67 years). Fifty-two (25%) patients were younger than 21 years, and 31 (15%) were older than 40 years, without a major difference compared with the distinct tumor entities.

Table 1. Seizure data and outcome in 207 patients with epilepsy-associated tumors
CharacteristicsEngel IEngel II–IVAll casesUnivariate analysis p Value
n(%)n(%)n(%)
  1. p values were estimated by chi-square test or Fisher's exact test dependent on the sample size.

All169823818207100 
Sex       
 Male8683181710450 
 Female83812019103500.6950
Age at first seizure13 13 13 (0–62) 0.8834
 [median yr (range)]       
Age at surgery28 33 28 (5–67) 0.0755
 [median yr (range)]       
Duration of epilepsy before surgery12 17 12 (2–54) 0.0669
 [median yr (range)]       
Secondary generalization       
 Yes10279272112962 
 No6786111478380.2188
Tumor location       
 Frontal20805202512 
 Temporomesial9685171511355 
 Temporolateral447713235728 
 Other9753251260.5110
Side       
 Left8982201810953 
 Right8082181898470.9972
Invasive evaluation       
 Yes417315275627 
 No128851415151730.5896
Additional EEG focus       
 Yes18679332713 
 No151842916180870.0311
Hippocampal sclerosis       
 Yes24036052 
 No388851243210.0029
Approach       
 2/3 Lobectomy607817227737 
 1/3 Lobectomy141000 147 
 Transsylvian3695253818 
 Temporodorsal1267633189 
 Transcortical4778132260290.0230
Hippocampal resection       
 Yes11582261814168 
 No5482181866320.9644
Amygdala resection       
 Yes698512158139 
 No100792621126610.2911
Histology (WHO grade)       
 Pilocytic astrocytoma I24739273316 
 Astrocytoma II257110293517 
 Astrocytoma III26713331 
 Oligodendroglioma II1387213157 
 Ganglioglioma I708512158240 
 Ganglioglioma II41000 42 
 Ganglioglioma III11000 11 
 Dysembryoplastic neuroepithelial tumor I25864142914 
 Pleomorphic xanthoastrocytoma II51000 550.4049
Astrocytoma (pilocytic, grade II or III)       
 Yes517220287134 
 No118871813136660.0084
WHO grade       
 I11983251714470 
 II478012205929 
 III375125420.8321
Residual tumor       
 Yes247110293416 
 No145842816173840.0686
Residual oligodendroglioma       
 Yes13326631 
 No121000 1260.0286

Preoperatively, 203 (97.5%) of the patients had complex partial seizures, in 62% combined with secondarily generalized tonic–clonic seizures. In 56 (27%) patients, the invasive evaluation with subdural grid or strip electrodes and/or intrahippocampal depth electrodes was necessary to determine the epileptic focus and to decide the extent of resection.

Presurgical EEG recordings revealed a focus additional to the site of the lesion in 13% of the patients. The side of the additional focus did not influence seizure outcome. Seven (70%) of the ten patients with EEG foci over the same side became seizure free, as did 11 (65%) of the 17 patients with foci over the opposite side. the tumors were found mainly in the temporomesial lobe (55%), and only 25 (12%) patients had with frontal location. Hippocampal sclerosis was identified from removed tissue in only four patients because of tissue alteration in temporomesial tumors, and in one, it was detected on MRI in the contralateral hemisphere. Additional cortical dysplasia was detected in four (2%) patients, and none of them became seizure free after operation.

One year after surgery, 169 (82%) patients were seizure free (Engel class I); seven (3%) had only rare seizures (class II), 20 (10%) had worthwhile improvement (class III), and 11 (5%) had no benefit. In the 38 patients with unfavorable seizure outcome (class II–IV) 1 year after surgery, 12 (32%) patients showed epileptic discharge in follow-up EEGs, but four of these patients became seizure free later. In 67 (40%) of the seizure-free patients, the antiepileptic drugs (AEDs) could be discontinued. Actuarial long-term analysis as year-to-year seizure outcome demonstrates a very stable percentage over an 11-year follow-up (Fig. 1). Epilepsy recurred in only 18 (11%) of the 169 primary seizure-free patients, with a maximum rate around the third and fourth year. Three of the 11 patients with later seizure cure had a second operation with resection of residual tumor in two and of cortical dysplasia in one patient.

Figure 1.

Year-to-year analysis of seizure outcome demonstrates a very stable percentage of seizure-free patients over a 10-year follow-up.

Twenty-nine percent (n = 11) of those 38 patients who were in outcome classes II–IV 1 year after surgery improved to outcome class I after a median follow-up of 3 years (1–6 years). Thus 14% of the group changed their seizure-outcome class during long-term follow-up.

The following factors improved seizure outcome 1 year after surgery in univariate analysis (Table 1): Shorter duration of epilepsy before surgery (without significance), single EEG focus, absence of additional hippocampal sclerosis or cortical dysplasia, transsylvian approach, other histology than Astros (pilocytic or WHO grade II or III), and no residual tumor tissue, especially in Oligos and DNTs.

The 27 patients with an additional EEG focus had a median duration of epilepsy before surgery of 16 years as opposed to 12 years of the whole group. Multifactorial analysis (Table 2) showing the odds ratios of these factors confirmed that a longer duration of preoperative epilepsy, an additional EEG focus, additional hippocampal sclerosis or cortical dysplasia, temporodorsal approach, Astros (pilocytic or WHO grade II or III), and incomplete tumor removal decreased the chance for seizure cure, whereas patients in whom the tumors could be reached over the transsylvian approach had a 4.6-fold chance for seizure freedom.

Table 2. Stepwise logistic regression analysis of influence factors on seizure outcome
CharacteristicsOdds ratio for Engel class I95% confidence limitsp-value
Duration of epilepsy before surgery decrease of risk per year0.9560.9210.9930.0205
Additional EEG-focus0.4020.1431.1330.0845
Additional hippocampal sclerosis0.1230.0121.2180.0732
Transsylvian approach4.6081.00821.2770.0488
Temporo-dorsal0.3340.1051.0580.0623
Astrocytoma (pilocytic, grade II or III)0.3370.1540.7380.0065
Residual tumor0.4150.1631.0580.0655

To examine changes of the operative procedures during the 12-year period, these were analyzed in four 3-year periods. Although the type of resection changed to smaller and more tailored types, seizure outcome improved from 69% to 84% in class I during the observation period (Fig. 2).

Figure 2.

The frequency of the use (ordinate) of five different resection types (axis of abscissa) as distributed over four time periods: The four different columns denote the frequency within each time period. The bottom line gives the percentage of class I outcome for each of the four time periods. Whereas the two-thirds resection was the main approach between 1988 and 1990 and decreased from >70% to 25%, the transsylvian and the transcortical resections were used more frequently. Throughout the four time periods, the seizure cure increased from 71% up to 88%. Temp, temporal; res, resection.

Tumor outcome and influence factors

Histology revealed 144 (70%) WHO grade I tumors (GGs, 82; pilocytic Astros, 33; DNTs, 29), 59 (29%) WHO grade II lesions (Astros, 35; Oligos, 15; pXAs, five; GGs, four), and four (1%) WHO grade III tumors (Astros, three; GGs, one) (Table 1). The demographic data like median age at first seizure, age at operation, duration of seizure disorder and follow-up, but also the location of the lesion did not show significant differences between the distinct tumor types.

All LEATs were found mainly in the temporal lobe (83%). Even Astros grade II or III showed a frontal location in only 21% (temporal, 68%). According to the postoperative MRI, incomplete tumor resection was found in 34 (16%) of the 207 patients, with a higher rate in Astros grade II and Oligos (20%). To remove both the tumor and the epileptogenic focus, the hippocampal structures were excised in 141 (68%) patients.

Ki-67 was found in between 0.5% and 2% in all grade I tumors beside three GGs and one DNT, which showed levels <5%. WHO grade II tumors showed Ki-67 values ≤5%, but none of the tumors, which later recurred, had higher levels of Ki-67 than the others. Focusing on the proliferation antigen Ki-67 in the astrocytomas grade II, a remarkable low rate (<1%) was found in 54% of the cases, including one with later recurrence.

No operative or postoperative mortality was found, with a transient morbidity of 6% (meningitis, two; pneumonia, one; thrombosis, four) with pulmonary embolism in two, nerve compression from malpositioning in one, secondary wound healing in one, infratentorial subarachnoidal hemorrhage in one, epidural hematoma in one, and frontal facial paresis in two. Permanent morbidity was seen in two patients: one has incomplete aphasia, and the other, hemiplegia. Nine of the 207 patients had a second operation, for residual tumor in four, for recurrent tumor in four, and for cortical dysplasia in one. In case of anaplastic tumors, all patients were advised to have additional radio- and chemotherapy, but one patient with astrocytoma and one with ganglioglioma, both operated on in1992, refused. Two other patients received additional radio- and chemotherapy after they showed inoperable tumor recurrence with malignant progression.

Only eight (4%) of the 207 patients showed tumor recurrence within a mean of 3 years after operation, but their initial tumors did not differ in the level of the proliferation antigen Ki-67 from the others without recurrence. The four (2%) patients with malignant progression died within a mean of 4 years after resection. Only two of the four anaplastic cases had recurrent tumors, and no one died. Thus the 8-year median survival was 98%, and the 8-year median recurrence was 3% in the 180 patients who had a minimal follow-up of 5 years. In the 38 patients with Astro grade II or III, the calculated 10-year survival by using the Kaplan–Meier method was 90%, and the calculated 10-year recurrence was 25% (Fig. 3), quite different from published series. Besides one patient with a GG grade I who died of malignant progression into glioblastoma 3 years after operation, no other patient with typically epilepsy-associated tumors (GG, DNT, pilocytic Astro, pXA) showed recurrence or death.

Figure 3.

Kaplan–Meier curves demonstrate a 10-year recurrence rate of 25% and a 10-year survival rate of 90% in 38 patients with Astro WHO grade II or III and long-term epilepsy.

Analyzing the clinical data for influence of recurrence and survival (Table 3), the following differences between patients with or without tumor progression could be found: Patients with recurrent tumors were older at the time of operation, and no one was younger than 20 years; 16% of the frontal tumors recurred, as opposed to only 3% of temporal tumors; the gemistocytic variant only present in one patient led to recurrence and death; and patients with residual tumor tissue had double the risk of recurrence and a ninefold risk of death.

Table 3. Tumor data and outcome in 207 patients with epilepsy associated tumors
 All casesRecurrenceDeath
Characteristicsn(%)n(%)n(%)
All2071008442
Sex      
 male104503311
 female103505533
Duration of epilepsy before surgery12 (2–54)9 (3–54)12 (4–30)
 [median years (range)]      
Age at surgery28 (5–67)47 (25–67)43 (24–53)
 [median years (range)]      
Time to recurrence/death3 (1–4)  4 (3–5)
 [median years (range)]      
Tumor location      
 frontal2512416312
 temporo-mesial113552211
 temporo-lateral5728240 
 other1260 0 
Histology (WHO grade)      
 pilocytic astrocytoma I33160 0 
 astrocytoma II351751439
   fibrillary13632318
   oligo-astrocytoma105110110
   gemistocytic1111001100
   no differentiation1151100 
 astrocytoma III312660 
 oligodendroglioma II1570 0 
 ganglioglioma I82401111
 ganglioglioma II420 0 
 ganglioglioma III100 0 
 dysembryoplastic neuroepithelial tumor I29140 0 
 pleomorphic xanthoastrocytoma II520 0 
WHO grade      
 I144701111
 II59295835
 III422500 
Residual tumor      
 yes34162639
 no173846311

DISCUSSION

Study design

This is a cohort study on 207 consecutive tumor cases obtained from a large series of epilepsy surgeries for drug-resistant epilepsy, looking simultaneously at the long-term seizure outcome and at the long-term oncologic aspects. This clinical trial differs in three aspects from previous reported series including our own (8,10,19,22,25,26,28,30,39): First, all patients had pre- and postoperative MRIs, which improve the diagnosis and the planning of the resection and define more exactly the extent of resection (40). Second, all biopsy patients were excluded, because the epileptogenic focus had to be removed, too. The last difference, compared with other tumor series, is the fact that except for the two patients with malignant tumor recurrence and for the two with anaplastic Astros, none had adjuvant radiotherapy or chemotherapy after the first surgery.

Seizure outcome

Several authors described factors that improve seizure outcome: complete resection of the lesion (5,27,28) and shorter duration of the preoperative seizure disorder (29). This series presents rare long-term data (median follow-up, 8 years) with complete cure from seizures in 82% as opposed to ∼70% in other groups (5,28,29). Our data showed that patients with a second EEG focus not only had a lower rate of seizure relief but also had a longer history of epilepsy that might have contributed to the second focus. Complete tumor resection improved seizure outcome in this series. In the series of Kirkpatrick et al. (25) with 87% DNTs, poorer seizure outcome was not associated with residual tumor tissue, but the extent of resection was determined only on CT scans. Age at operation in the present study did not influence the rate of seizure cure in contrast to Morris et al. (29). According with Morris, the age at seizure onset did not correlate with the epilepsy outcome, and shorter duration of preoperative epilepsy was associated with better seizure outcome corresponding to our decreased chance of seizure cure per year of present epilepsy (29).

In addition to the current literature, we found that the tumor type correlated with seizure outcome. Thus patients with GGs and Oligos became seizure free in >90%, as opposed to patients with Astros II (only 66%) and to patients with pilocytic Astros (only 61%). Lower rates of seizure cure in these Astros could be due to the following factors: (a) higher rate of residual tumor, and (b) significantly longer duration of preoperative epilepsy, especially in patients with pilocytic Astro I. A remarkable finding is that the overall seizure outcome over 11 years remained stable. It is well known that seizure outcome slowly deteriorates in the long-term follow-up, with a rate of ∼1–1.5% loss in the class I outcome group per year. This corresponds to 8% of primary class I outcome deterioration over the next 5 years in this group or 11% over 13 years. However, remarkably, 11 of 38 primary non–seizure-free patients became seizure free during follow-up in the second to fifth year, representing one third of these patients.

Surgical techniques

Since the beginning of this epilepsy surgery program, the extent of resection, especially of the temporal lobe has decreased, presumably because of more exact focus evaluation and better MRI demarcation of the lesion, certainly due to growing experience, and because of new principles. During the same period, seizure outcome has improved. This also seems to suggest that in tumoral epilepsy surgery, it is safe to go for a more restricted resection instead of the safer-appearing standard anterior 2/3 resection, a finding also confirmed in our series on neocortical temporal lobe epilepsy, and as suggested by Spencer and Fried (2,27,30).

Tumor outcome

Tumor outcome in 144 patients with GGs, DNTs, and pilocytic Astros and a median follow-up of 8 years was excellent, with only one recurrence in GG WHO grade I, which showed malignant transformation and death 3 years after initial surgery. Even patients with GG WHO grade II showed no recurrence. This benign course of GGs, DNTs and pilocytic Astros corresponds to those in other series (6–8,10,11). Only one GG study reported a higher recurrence rate of 25% (41).

In patients with supratentorial WHO grade II or grade III Astros and intractable epilepsy, our data demonstrate remarkable findings: a high 10-year survival rate of 90% and a low 10-year recurrence rate of only 25%. This is in contrast to other grade II or III Astro series without long-term epilepsy, showing a 5-year survival rate of 50% (13), 64% (16), 60% (17), and a 10-year survival rate of 21% (24). Some factors are known to be associated with a lower recurrence rate and a longer survival rate in grade II or III Astros, and they were found to be present in this series: (a) temporal location in 71% as opposed to 30%, (b) low rate of frontal location in only 23% against 30% and 43%, (c) lower age at the operation with a peak incidence at the age of 20 to 30 years as opposed to those between 30 and 40 years, (d) the gemistocytic variant in only 0.5% compared with 10%, (e) gross total resection in 83% compared with only 19%–50%, and (f) mean duration of preoperative symptoms of 15 years compared with ∼1 year (13,15–17,20,22,24).

Piepmeier et al. (21) suggested that in patients with low-grade gliomas, the overall time between the onset of symptoms and death remains constant and that the seemingly longer survival was only a consequence of earlier diagnosis and operation. For the LEAT subgroup of grade II or III Astros, this does not appear to be true: They have a preoperative average history of 12 years and a mean for the total history (first seizure to end of follow-up) of 21 years, very close to the mean for the total history of 22 years for GGs and 20 years for DNTs. In contrast to this, duration of symptoms for non-LEAT Astros, WHO grade II, is usually between 1 and 2 years. Adding a mean survival rate of 5 years, a total duration of history of 6.5 years leads to a median survival rate between 50 and 65% at a maximum at those with 5 years. In the LEAT Astros grade II or III, the total duration of history at this point would be 20 years, with a 90% survival rate. These differences between long-term epilepsy–associated Astros grade II or III and non-LEAT Astros may make a thorough review of this LEAT subgroup necessary by using refined and more modern histologic methods as well as further long-term follow-up.

We know of only one other group that is aware of this particular behavior of a certain subgroups of LEAT patients. Piepmeier et al. (42) reported a 10-year recurrence-free survival rate of 100%, regardless of the extent of surgery. Could it be that seizures themselves are the prognostic indicator for the extremely benign course in these grade II or III LEATs? Seizures per se appear to be an unlikely indicator, because ∼90% of patients with non-LEAT astrocytic lineage tumors have seizures as their first symptom, but the 5-year survival rate in these non-LEAT Astros rarely reaches >65%. Thus the Astro grade II or III subgroup has clinical features in common with other LEAT subgroups (such as GGs) but gave the histologic appearance of grade II or III astrocytic tumors, and it may not be easy to explain these inconsistent findings.

A future task will be to analyze factors in this interesting LEAT subgroup, to find out special histopathologic features that may be indicative of a more favorable outcome, if they do exist. The analysis of all LEATs has confirmed the well-known good prognosis of GGs and DNTs, but suggested that a subgroup of grade II Astros may be hidden among the LEATs, which are presumed to have a different tumor biology. Additionally this LEAT subgroup will have to be reviewed neuropathologically and followed up clinically for several more years.

Conclusions

Seizure aspects

Patients with epilepsy for >2 years (mean duration of 15 years) and with neuroepithelial tumors reached long-term seizure relief in 80% according to Engel's outcome scale, and in 41% without AEDs. No mortality and only 1% permanent morbidity were noted. High rates of seizure relief are maintained despite a change to more-restricted resection types. Factors associated with poorer seizure outcome were longer duration of epilepsy before surgery, additional EEG focus, additional hippocampal sclerosis or cortical dysplasia, Astros, and incomplete tumor resection. Improvement of seizure outcome is possible when the tumor is diagnosed and removed early and completely.

Tumor aspects

Patients with drug-resistant epilepsy for >2 years before the operation and WHO grade II or III Astros have an uncommonly long 10-year survival rate of 90% with a unusually low 10-year recurrence of only 25%. Despite a long history, tumor recurrence with malignant progression and death occurs in some patients, also with GGs; therefore tumor removal is advisable. Age at operation older than 40 years, frontal tumor location, gemistocytic differentiation, and incomplete resection are associated with poorer tumor outcome. Astros grade II or III prospective long-term survivors cannot be distinguished by the histopathologic findings from tumors with early and malignant progression, so that even in patients with a long duration of symptoms, regular MRI control is indispensable. Further investigations are necessary to identify prognostic factors in these patients at the time of operation.

Acknowledgments

Acknowledgment: The study was supported by grants (SFB 400, TP B1; SFB TR 3, TP A1, and Schr 285-1/2) from the Deutsche Forschungsgemeinschaft.

Ancillary