Abstract
Objective
To quantify size and localization differences between tumors presenting with seizures vs nonseizure neurological symptoms.
Design
Retrospective imaging survey. We performed magnetic resonance imaging–based morphometric analysis and nonparametric mapping in patients with brain tumors.
Setting
University-affiliated teaching hospital.
Patients or Other Participants
One hundred twenty-four patients with newly diagnosed supratentorial glial tumors.
Main Outcome Measures
Volumetric and mapping methods were used to evaluate differences in size and location of the tumors in patients who presented with seizures as compared with patients who presented with other symptoms.
Results
In high-grade gliomas, tumors presenting with seizures were smaller than tumors presenting with other neurological symptoms, whereas in low-grade gliomas, tumors presenting with seizures were larger. Tumor location maps revealed that in high-grade gliomas, deep-seated tumors in the pericallosal regions were more likely to present with nonseizure neurological symptoms. In low-grade gliomas, tumors of the temporal lobe as well as the insular region were more likely to present with seizures.
Conclusions
The influence of size and location of the tumors on their propensity to cause seizures varies with the grade of the tumor. In high-grade gliomas, rapidly growing tumors, particularly those situated in deeper structures, present with non–seizure-related symptoms. In low-grade gliomas, lesions in the temporal lobe or the insula grow large without other symptoms and eventually cause seizures. Quantitative image analysis allows for the mapping of regions in each group that are more or less susceptible to seizures.
Seizures are encountered in a majority of patients with primary brain tumors and are a major cause of morbidity in these patients.1,2 Thirty percent to 50% of patients experience a seizure by the time their tumors are diagnosed, and an additional 6% to 45% of patients who do not initially present with seizures eventually develop them.3-5 Characteristics of brain tumors and their mechanism in causing seizures in patients are incompletely understood.4,6 Low-grade, well-differentiated gliomas,1,6-9 cortically located tumors,3,10-14 and location in the temporal/frontal and motor/sensory cortices6,8,15-17 are more frequently associated with seizures.
Although there is a high incidence of seizures in these patients, treatment strategies remain poorly defined. Prophylactic anticonvulsant therapy, shown to be ineffective in preventing seizures in patients with brain tumors in multiple large-scale studies,12,18-20 is not recommended by the American Academy of Neurology.5 None-theless, prophylaxis remains a wide-spread practice21 because of difficulty in determining which patients are at greatest risk for seizures. Determination of morphometric factors influencing seizures would help in identifying patients at greatest risk for early, targeted treatment and prevent potentially toxic, unnecessary treatment in patients at minimal risk.
Although studies examining brain tumors in relationship to epilepsy have localized tumors to a particular lobe,10 few studies have performed quantitative volumetric or spatial mapping analysis of tumors in relation to their epileptogenic potential. Regions within a particular lobe are likely to exhibit different epileptogenic potential to tumor invasion and tumors frequently affect multiple contiguous lobes.6
Modern imaging techniques allow for analysis of lesions over a large group of subjects through registration and mapping techniques. In this study, we used these techniques to examine the size and location of primary supratentorial glial brain tumors and characterized their propensity to cause seizures at presentation.
METHODS
This retrospective study examined patients who underwent surgical evaluation of a brain tumor at the Brigham and Women’s Hospital between January 2005 and September 2007. Inclusion criteria were age 18 years or older, new diagnosis of brain tumor, supratentorial location, pathologically proven glial tumor, and preoperative acquisition of high-quality volumetric magnetic resonance imaging (MRI) scan. Tumors were designated “low grade” (World Health Organization grade I or II) or “high grade” (World Health Organization grade III or IV).
A total of 81 patients presented with low-grade gliomas during the study period. Of these, 24 patients were excluded from the analysis because of lack of access to preoperative volumetric MRI scans (13 patients), unclear clinical history (2 patients), clinical or radiological suspicion of higher-grade lesion than suggested from pathological examination (3 patients), poorly defined lesion (1 patient), suspicion of neuroglial component (1 patient), and inconclusive pathological examination (6 patients).
All 57 patients with low-grade glioma who met the criteria were analyzed. Because of their vastly larger numbers, 67 consecutive patients with high-grade gliomas who obtained identical MRI sequences between the dates March 24, 2005, and May 20, 2006, from 317 eligible patients during the study period, were selected. Patient records were reviewed to determine the presenting symptom. Approval for this study was obtained from the local human research institutional review board.
IMAGE ACQUISITION
Patients with low-grade tumors underwent preoperative imaging with 1 of several MRI scanners from which T2 and volumetric T1 images were obtained: 0.5-T MRI (SignaSP; GE Medical Systems, Milwaukee, Wisconsin); 1.5-T General Electric Signa Excite scanner; and 3-T General Electric Signa scanner. Patients with high-grade tumors underwent imaging with a 1.5-T General Electric Signa scanner.
IMAGE PROCESSING
Tumors were manually segmented from the MRI by a blinded rater using standard image processing software to create a lesion mask (3D Slicer; www.slicer.org and MNI Display; www .bic.mni.mcgill.ca). For low-grade tumors, T2-weighted images were resampled and registered into T1-weighted space; the tumor margin was determined by the extent of T2 signal abnormality.22,23 For high-grade gliomas, tumor margin was delineated by the area of contrast enhancement on the volumetric T1-weighted image.24
Magnetic resonance images were transformed into a standardized coordinate space based on the Talairach atlas25 to account for differences in brain orientation and differences in intracranial volume. Automatic registration using linear affine transformation was performed from the T1-weighted images.26 Because distortions of the anatomy caused the registration procedure to fail at times, registration validity was checked by selecting 6 points on both the template and the target brains (maximal anterior and posterior cortical extent along the anterior-posterior commissure line, upper and lower extent along the perpendicular line through the anterior commissure, left and right extent along the third axis formed by the 2 previous lines); if the root mean square was greater than 5 mm, registration was reperformed using the manually selected coordinates (MNI Register; www.bic.mni.mcgill.ca). Tumor volumes were calculated after registration (MATLAB; MathWorks, Natick, Massachusetts).
STATISTICAL ANALYSES
Logistic regression was used to estimate odds ratios and 95% confidence intervals for patient characteristics. SAS version 9.1 was used (SAS Institute, Cary, North Carolina). To assess the localizing value of the tumors causing seizures, a χ2 statistic map was calculated. At each voxel, the group of patients presenting with a tumor at that voxel was determined. From this group, the number of patients who presented with seizures, as compared with the expected number (the product of the number of patients with a tumor at that voxel and the ratio of the patients presenting with seizures in the study population), was used to calculate the χ2 statistic. In voxels where the number of patients with seizures exceeded the number of patients without seizures, a higher value of χ2 is indicative of a stronger likelihood that patients with a tumor at that location would present with seizures than at other locations. A complementary χ2 map was calculated to determine “protective” locations, eg, where presentation with seizures is less likely.
To assess the significance of the χ2 statistic, a nonparametric mapping method based on permutation testing was used,27 which requires making minimal assumptions regarding our data. The χ2 was used as an omnibus statistic. The χ2 map was masked to include only voxels that exceeded the suprathreshold level of 2. These were grouped into discrete clusters using a 6-connectivity model, and a cluster mass was calculated.28 Thereafter, the labeling of each patient as “seizure” vs “no seizure” was randomly reassigned, with the constraint of preserving the original ratio of the number of patients with “seizure” and “no seizure.” From the relabeled group, a χ2 map and clusters were again calculated and the largest cluster mass recorded. This was repeated 5000 times to obtain a null hypothesis distribution of maximal cluster sizes.28,29 Clusters of the original image whose masses exceeded significance of P<.05 were used to calculate a P value map. Analysis was performed using MATLAB version 7.6 (R2008a).
RESULTS
PATIENT POPULATION
One hundred twenty-four patients were included in this study. Their main clinical and pathological characteristics are listed in Table 1; odds ratios regarding these characteristics are listed in Table 2. Patients with presumed low-grade gliomas not included in the study were compared with patients who were included. Of the 24 patients, 15 were male, 14 had left-sided tumors, 7 had temporal tumors, and 15 had seizures; their average age was 46.3 years (range, 18-83 years). The age of the excluded patients was higher than the included group, but the other characteristics did not differ significantly.
Table 1. Patient Demographics.
| No. of Patients |
||
|---|---|---|
| Seizure (n = 61) |
No Seizure (n = 63) |
|
| High grade | 26 | 41 |
| Low grade | 35 | 22 |
| Histology | ||
| Glioblastoma | 16 | 33 |
| Astrocytoma | 14 | 9 |
| Oligodendroglioma | 15 | 13 |
| Oligoastrocytoma (mixed) | 16 | 8 |
| Age, y, median (range) | 42 (18-88) | 50 (18-88) |
| High grade | 55 (26-88) | 57 (21-88) |
| Low grade | 38 (18-56) | 38 (18-64) |
| Sex, M/F | 33/28 | 28/35 |
| High grade | 16/10 | 23/18 |
| Low grade | 17/18 | 5/17 |
| Hemisphere, L/R | 32/29 | 37/24 |
| High grade | 15/11 | 25/14 (2 midline) |
| Low grade | 17/18 | 12/10 |
| Temporal | 31 | 18 |
| Extratemporal | 30 | 45 |
| High grade, temporal | 15 | 14 |
| High grade, extratemporal | 11 | 27 |
| Low grade, temporal | 16 | 4 |
| Low grade, extratemporal | 19 | 18 |
| Location, cortical/noncortical | 51/2 | 40/19 |
| High grade | 22/1 | 21/18 |
| Low grade | 29/1 | 19/1 |
| Hippocampal location | 8 | 4 |
| High grade | 3 | 2 |
| Low grade | 5 | 2 |
| Tumor volume, cm3, median | 34.7 | 39.7 |
| High grade | 30 | 46.7 |
| Low grade | 37.8 | 14.4 |
| High-grade glioma: left | 23.7 | 46.7 |
| High-grade glioma: right | 34.7 | 45.7 |
| Low-grade glioma: left | 48.4 | 15.1 |
| Low-grade glioma: right | 34 | 13.5 |
Table 2. Odds Ratios and 95% Confidence Intervals for Patient and Tumor Characteristics to Predict Presentation With Seizures.
| No. (%) |
||||
|---|---|---|---|---|
| Variable | No Seizure (n=63)a |
Seizure (n=61)a |
Odds Ratio (95% Confidence Interval) |
P Value |
| Grade | .01 | |||
| Low | 22 (35) | 35 (57) | 2.51 (1.22-5.19) | |
| High | 41 (65) | 26 (43) | 0.40 (0.19-0.82) | |
| Age per decade, y, median (range) | ||||
| Overall | 50 (18-88) | 42 (18-88) | 0.77 (0.61-0.96) | .02 |
| Low grade | 38 (18-64) | 38 (18-56) | 1.00 (0.60-1.66) | .99 |
| High grade | 57 (21-88) | 55 (26-88) | 0.80 (0.59-1.10) | .17 |
| Tumor volume, median (range), per 10 cm3a | ||||
| Overall | 41 (1.0-166) | 35 (0.3-205) | 1.05 (0.96-1.15) | .31 |
| Low grade | 14 (1.0-87) | 38 (0.3-205) | 1.31 (1.06-1.60) | .01 |
| High grade | 47 (2.6-167) | 30 (2.0-143) | 0.88 (0.74-1.04) | .13 |
| Male | ||||
| Overall | 28 (44) | 33 (54) | 1.47 (0.73-2.99) | .28 |
| Low grade | 5 (23) | 17 (49) | 3.21 (0.97-10.63) | .06 |
| High grade | 23 (56) | 16 (62) | 1.25 (0.46-3.41) | .66 |
| Temporal | ||||
| Overall | 18 (29) | 31 (51) | 2.58 (1.23-5.43) | .01 |
| Low grade | 4 (18) | 16 (46) | 3.79 (1.06-13.51) | .04 |
| High grade | 14 (34) | 15 (58) | 2.63 (0.96-7.23) | .06 |
| Hippocampal | ||||
| Overall | 4 (6.4) | 8 (13.1) | 2.23 (0.63-7.82) | .21 |
| Low grade | 2 (9.1) | 5 (14.3) | 1.67 (0.29-9.44) | .56 |
| High grade | 2 (4.9) | 3 (11.5) | 2.54 (0.40-16.37) | .32 |
| Corticalb | ||||
| Overall | 40 (63) | 51 (84) | 2.93 (1.25-6.86) | .01 |
| Low grade | 19 (86) | 29 (83) | 0.76 (0.17-3.43) | .72 |
| High grade | 21 (51) | 22 (85) | 5.24 (1.53-17.91) | .01 |
| Left hemisphere | ||||
| Overall | 37 (59) | 32 (52) | 0.78 (0.38-1.58) | .48 |
| Low grade | 12 (55) | 17 (49) | 0.79 (0.27-2.29) | .66 |
| High grade | 25 (61) | 15 (58) | 0.87 (0.32-2.37) | .79 |
Interaction with grade, P=.004.
Interaction with grade, P=.05.
Patients with low-grade tumors were more likely to present with seizures than patients with high-grade tumors. Age was significantly protective overall (P=.02) with a reduction in likelihood of 23% per decade. However, subgroup analysis revealed that patients with high-grade tumors were significantly older than those with low-grade tumors.
LOBE INVOLVEMENT
Patients with tumors involving the temporal lobe were more than twice as likely to present with seizures. This tendency was significant in low-grade tumors and there was a similar but slightly weaker tendency in patients with high-grade tumors. Low-grade tumors in the temporal lobe were more likely to present with seizures (16 of 20 patients). High-grade tumors in the temporal lobe were neither more nor less likely to present with seizures (14 of 29 patients).
VOLUME AND SEIZURES
There was significant qualitative interaction between tumor volume and grade (P=.004). Tumor volume was predictive of presenting with seizures in patients with low-grade tumors (P=.01), with a 3% increase in likelihood per cubic centimeter (Figure 1). In high-grade tumors, there was a slight tendency for tumor volume to be protective in relation to seizure presentation (P=.13).
Figure 1.
Tumor volume and grade.
LOCATION MODELING
The aggregate of all the tumors generated from the sum of the binary tumor masks is shown for high-grade (Figure 2A) and low-grade (Figure 2B) gliomas. The χ2 maps of high-grade tumors were created indicating where seizures were more likely to occur and where other neurological symptoms were more likely to occur. Cluster analysis revealed 1 cluster that reached statistical significance (P<.05) to indicate regions where patients were more likely to present with nonseizure neurological symptoms (Figure 3). The cluster was located in the frontal pericallosal region and was 70.4 cm3 in volume. No clusters of significant size were found indicating where seizures were more likely to occur, including the temporal region.
Figure 2.
Summed statistic image. At each voxel, the number of patients presenting with tumors is calculated. A, High-grade tumors. B, Low-grade tumors.
Figure 3.
High-grade tumors χ2 map. A χ2 value was calculated at each voxel from the expected number and the observed number of tumors at that voxel. Region of statistical significance (P<.05) where nonseizure neurological symptoms are more likely was calculated by clustering analysis.
The χ2 maps of low-grade tumors indicating where seizures were more likely to occur and where other neurological symptoms were more likely to occur were created. Significant clusters where patients were more likely to present with seizures were found in the right hemisphere (size 56.5 cm3) and in the left hemisphere (size 70.4 cm3) (Figure 4A). Clusters in both hemispheres involved the temporal lobe, while the right hemisphere cluster included the insular region. There were no statistically significant clusters to denote regions where patients were more likely to present with nonseizure neurological symptoms.
Figure 4.
Low-grade tumors χ2 map. A, Region of significance where seizures are more likely. B, Analysis repeated after inverting all right temporal tumors to the left hemisphere.
Because of concern of uneven spatial distribution of the low-grade tumors between the 2 hemispheres, cluster analysis was repeated after inverting all right hemispheric tumors across the midline to simulate 58 left hemispheric tumors. This revealed a single significant cluster of 115.6 cm3 that included the insular as well as most of the temporal lobe (Figure 4B).
SEIZURE SEMIOLOGY
Of the 26 patients with high-grade tumors with seizures, 11 patients had generalized convulsions; 7 had simple partial seizures; and 7 had complex partial seizures. There was insufficient information to determine accurate seizure semiology in 1 patient. Of the 35 patients with low-grade tumors and seizure presentations, 21 had generalized convulsions; 10 had simple partial seizures; and 4 had complex partial seizures. No significant differences were seen between the 2 groups (P=.24).
COMMENT
We examined the effects of glioma size and location on propensity to generate seizures as an early symptom using quantitative image analysis techniques.
SEIZURES AND SIZE OF TUMOR
In this study, 34% of patients with high-grade tumors and 62% of those with low-grade tumors presented with seizures, consistent with the previously reported values.1-3,8,30-36 High- and low-grade tumors differ in terms of their relationship between seizure size/location and the propensity to present with seizures. High-grade tumors presenting with seizures are likely to be smaller than those presenting with other symptoms. We postulate that rapidly growing tumors cause symptoms related to mass effect, such as headache, cognitive deficits, or focal weakness, rather than seizures. The reverse was found with low-grade tumors, where large tumors were more likely to present with seizures than small tumors. We postulate that large tumor size is indicative of the long duration of silent growth, allowing more time for seizures to develop.
Small low-grade tumors are sometimes found on imaging studies acquired to evaluate nonspecific symptoms that may be unrelated, as was the case in 7 patients. Whether tumors were completely incidental or whether they contributed to the presenting symptom is difficult to determine.
LOCATION OF THE TUMOR
Patients with high-grade tumors in the pericallosal region, likely representing so-called butterfly gliomas, were significantly less likely to present with seizures. Such deeply seated, rapidly growing tumors are more likely to cause symptoms due to mass effect rather than seizures.
Patients who presented with high-grade tumors in the temporal lobes were not more likely to present with seizures than other neurological symptoms. However, in comparison with the relatively low rates of seizures in deeply seated pericallosal tumors, this result indicates a relative increase in the epileptogenicity in this region. In contrast, patients with low-grade tumors were more likely to present with seizures if their tumors were located in the temporal lobe.
Patients with tumors in the insular cortex were also more likely to present with seizures. The insular cortex is often a region of seizure spread in temporal lobe epilepsy. Clinically, seizures originating from the insular region are difficult to distinguish from those arising from elsewhere in the temporal lobe37 and 10% of patients initially thought to have temporal lobe epilepsy may in fact have seizures originating in the insular cortex,38 though detailed investigation revealed distinct clinical differences.39,40 In a recent large retrospective review of 51 insular grade II gliomas, 50 patients presented with seizures and 45 patients had normal neurological examination results.41 High frequency of preoperative seizures has been reported in other smaller series describing low-grade insular tumors.3,42-45 These results suggest that tumors located in the insular cortex are likely to be clinically silent until the patient experiences a seizure.
ASYMMETRY OF LOW-GRADE TUMORS
Although the number of patients with low-grade tumors in the left hemisphere was equal to the number of right hemisphere tumors, and the volumes of the tumors in both hemispheres were comparable, the aggregate tumor image (Figure 2B) revealed that the distribution of the tumors was not identical; a large number of tumors were located in the right insular region, many of which presented with seizures. We are unaware of any studies in the literature that have systematically examined this asymmetry, though similar asymmetries have been found in some studies.41,43
Although volumes of regions of significance were similar over both hemispheres, regions of highest correlation with seizure presentation were located over the right hemisphere. It is possible that left hemisphere tumors are less likely to present with seizures, perhaps because of increased presentation with other neurological findings as a result of greater eloquence of the dominant hemisphere.
LIMITATIONS
Because of the retrospective nature of this study, we were unable to control for a number of factors. Magnetic resonance imaging scanning parameters were not homogeneous over the duration of this study. This is a historical artifact representing the development of imaging technology at our institution. We minimized the influence of this variability by selecting only patients who had received volumetric T1-weighted scans to obtain the highest-resolution images available at the time.
No regions of significantly increased risk for seizures were found for low-grade tumors in other areas where small tumors are believed to typically cause seizures (motor/sensory cortices, hippocampus). We also did not find that occipital tumors are unlikely to present with seizures. These negative findings may be due to a low number of tumors at those locations in our study.
It is likely that the tumor extends beyond the edge of the enhancing lesion in high-grade gliomas and that T2 signal changes in low-grade gliomas in some areas may represent an edematous process rather than tumor. However, these methods result in boundaries that are easily identified and provide reasonable delineation of tumors.
The registration procedure in patients with large lesions is challenging,46 causing automated registration to fail when tumors substantially distort cortical structure. We chose to correct the registration errors manually, which decreases interrater reproducibility as compared with automated techniques; such automated registration procedures are not yet well established.
We used a clustering algorithm rather than a voxel-based statistic. The cluster mass test is known to have increased sensitivity and specificity compared with tests based on voxel intensity when the signal is spatially extended,47-49 and as a result, these tests are more powerful than single-threshold approaches but result in reduced localizing power. We chose a low primary threshold level (χ2=2), resulting in large clusters emphasizing the high spatial correlation between adjacent voxels in our population; in doing so, we were unable to detect intense focal signals. The optimal selection of primary thresholds with permutation labeling remains to be resolved.27
Despite these limitations, volumetric imaging analysis is useful to localize regions that are particularly susceptible to seizures, as well as define regions that are relatively protective against seizures. Future studies using similar techniques may be used to identify patients who are at greater and lower risk for seizures prospectively and postoperatively, allowing for early selection of patients for targeted antiepileptic drug therapy.
Acknowledgments
Funding/Support: This study was supported by funding from the National Epifellows Foundation (Dr Lee), grant U41 RR019703 from the National Center for Image-Guided Therapy, and grants 5P01CA067165-11 and P41RR13218 from the National Institutes of Health.
Footnotes
Financial Disclosure: None reported.
REFERENCES
- 1.Hildebrand J, Lecaille C, Perennes J, Delattre JY. Epileptic seizures during follow-up of patients treated for primary brain tumors. Neurology. 2005;65(2):212–215. doi: 10.1212/01.wnl.0000168903.09277.8f. [DOI] [PubMed] [Google Scholar]
- 2.Moots PL, Maciunas RJ, Eisert DR, Parker RA, Laporte K, Abou-Khalil B. The course of seizure disorders in patients with malignant gliomas. Arch Neurol. 1995;52(7):717–724. doi: 10.1001/archneur.1995.00540310091021. [DOI] [PubMed] [Google Scholar]
- 3.Chang EF, Potts MB, Keles GE, et al. Seizure characteristics and control following resection in 332 patients with low-grade gliomas. J Neurosurg. 2008;108(2):227–235. doi: 10.3171/JNS/2008/108/2/0227. [DOI] [PubMed] [Google Scholar]
- 4.van Breemen MS, Wilms EB, Vecht CJ. Epilepsy in patients with brain tumours: epidemiology, mechanisms, and management. Lancet Neurol. 2007;6(5):421–430. doi: 10.1016/S1474-4422(07)70103-5. [DOI] [PubMed] [Google Scholar]
- 5.Glantz MJ, Cole BF, Forsyth PA, et al. Report of the Quality Standards Subcommittee of the American Academy of Neurology. Practice parameter: anticonvulsant prophylaxis in patients with newly diagnosed brain tumors. Neurology. 2000;54(10):1886–1893. doi: 10.1212/wnl.54.10.1886. [DOI] [PubMed] [Google Scholar]
- 6.Liigant A, Haldre S, Oun A, et al. Seizure disorders in patients with brain tumors. Eur Neurol. 2001;45(1):46–51. doi: 10.1159/000052089. [DOI] [PubMed] [Google Scholar]
- 7.Vertosick FT, Jr, Selker RG, Arena VC. Survival of patients with well-differentiated astrocytoma diagnosed in the era of computed tomography. Neurosurgery. 1991;28(4):496–501. doi: 10.1097/00006123-199104000-00002. [DOI] [PubMed] [Google Scholar]
- 8.Lynam LM, Lyons MK, Drazkowski JF, et al. Frequency of seizures in patients with newly diagnosed brain tumors: a retrospective review. Clin Neurol Neurosurg. 2007;109(7):634–638. doi: 10.1016/j.clineuro.2007.05.017. [DOI] [PubMed] [Google Scholar]
- 9.Lote K, Stenwig AE, Skullerud K, Hirschberg H. Prevalence and prognostic significance of epilepsy in patients with gliomas. Eur J Cancer. 1998;34(1):98–102. doi: 10.1016/s0959-8049(97)00374-2. [DOI] [PubMed] [Google Scholar]
- 10.Fried I, Kim JH, Spencer DD. Limbic and neocortical gliomas associated with intractable seizures: a distinct clinicopathological group. Neurosurgery. 1994;34(5):815–823. doi: 10.1227/00006123-199405000-00005. [DOI] [PubMed] [Google Scholar]
- 11.Bronen RA, Fulbright RK, Spencer DD, Spencer SS, Kim JH, Lange RC. MR characteristics of neoplasms and vascular malformations associated with epilepsy. Magn Reson Imaging. 1995;13(8):1153–1162. doi: 10.1016/0730-725x(95)02026-p. [DOI] [PubMed] [Google Scholar]
- 12.Glantz MJ, Cole BF, Friedberg MH, et al. A randomized, blinded, placebocontrolled trial of divalproex sodium prophylaxis in adults with newly diagnosed brain tumors. Neurology. 1996;46(4):985–991. doi: 10.1212/wnl.46.4.985. [DOI] [PubMed] [Google Scholar]
- 13.Penfield W, Erickson TC, Tarlov I. Relation of intracranial tumors and symptomatic epilepsy. Arch Neurol Psychiatry. 1940;44(2):300–315. [Google Scholar]
- 14.Smith DF, Hutton JL, Sandemann D, et al. The prognosis of primary intracerebral tumours presenting with epilepsy: the outcome of medical and surgical management. J Neurol Neurosurg Psychiatry. 1991;54(10):915–920. doi: 10.1136/jnnp.54.10.915. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Lund M. Epilepsy in association with intracranial tumour. Acta Psychiatr Neurol Scand Suppl. 1952;81:1–149. [PubMed] [Google Scholar]
- 16.Zaatreh MM, Spencer DD, Thompson JL, et al. Frontal lobe tumoral epilepsy: clinical, neurophysiologic features and predictors of surgical outcome. Epilepsia. 2002;43(7):727–733. doi: 10.1046/j.1528-1157.2002.39501.x. [DOI] [PubMed] [Google Scholar]
- 17.Zaatreh MM, Firlik KS, Spencer DD, Spencer SS. Temporal lobe tumoral epilepsy: characteristics and predictors of surgical outcome. Neurology. 2003;61(5):636–641. doi: 10.1212/01.wnl.0000079374.78589.1b. [DOI] [PubMed] [Google Scholar]
- 18.Forsyth PA, Weaver S, Fulton D, et al. Prophylactic anticonvulsants in patients with brain tumour. Can J Neurol Sci. 2003;30(2):106–112. doi: 10.1017/s0317167100053361. [DOI] [PubMed] [Google Scholar]
- 19.Franceschetti S, Binelli S, Casazza M, et al. Influence of surgery and antiepileptic drugs on seizures symptomatic of cerebral tumours. Acta Neurochir (Wien) 1990;103(1-2):47–51. doi: 10.1007/BF01420191. [DOI] [PubMed] [Google Scholar]
- 20.North JB, Penhall RK, Hanieh A, Frewin DB, Taylor WB. Phenytoin and postoperative epilepsy: a double-blind study. J Neurosurg. 1983;58(5):672–677. doi: 10.3171/jns.1983.58.5.0672. [DOI] [PubMed] [Google Scholar]
- 21.Siomin V, Angelov L, Li L, Vogelbaum MA. Results of a survey of neurosurgical practice patterns regarding the prophylactic use of anti-epilepsy drugs in patients with brain tumors. J Neurooncol. 2005;74(2):211–215. doi: 10.1007/s11060-004-6912-4. [DOI] [PubMed] [Google Scholar]
- 22.Talos IF, Zou KH, Ohno-Machado L, et al. Supratentorial low-grade glioma resectability: statistical predictive analysis based on anatomic MR features and tumor characteristics. Radiology. 2006;239(2):506–513. doi: 10.1148/radiol.2392050661. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Nimsky C, Fujita A, Ganslandt O, Von Keller B, Fahlbusch R. Volumetric assessment of glioma removal by intraoperative high-field magnetic resonance imaging. Neurosurgery. 2004;55(2):358–370. doi: 10.1227/01.neu.0000129694.64671.91. discussion 370-371. [DOI] [PubMed] [Google Scholar]
- 24.Lacroix M, Abi-Said D, Fourney DR, et al. A multivariate analysis of 416 patients with glioblastoma multiforme: prognosis, extent of resection, and survival. J Neurosurg. 2001;95(2):190–198. doi: 10.3171/jns.2001.95.2.0190. [DOI] [PubMed] [Google Scholar]
- 25.Talairach J, Tournoux P. Co-planar Stereotaxic Atlas of the Human Brain: 3-Dimensional Proportional System: An Approach to Cerebral Imaging. Thieme; Stuttgart, German: 1988. [Google Scholar]
- 26.Collins DL, Neelin P, Peters TM, Evans AC. Automatic 3D intersubject registration of MR volumetric data in standardized Talairach space. J Comput Assist Tomogr. 1994;18(2):192–205. [PubMed] [Google Scholar]
- 27.Nichols TE, Holmes AP. Nonparametric permutation tests for functional neuroimaging: a primer with examples. Hum Brain Mapp. 2002;15(1):1–25. doi: 10.1002/hbm.1058. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Bullmore ET, Suckling J, Overmeyer S, Rabe-Hesketh S, Taylor E, Brammer MJ. Global, voxel, and cluster tests, by theory and permutation, for a difference between two groups of structural MR images of the brain. IEEE Trans Med Imaging. 1999;18(1):32–42. doi: 10.1109/42.750253. [DOI] [PubMed] [Google Scholar]
- 29.Manly BJF. Randomization and Monte Carlo Methods in Biology. Chapman and Hall; London, England: 1991. [Google Scholar]
- 30.Ulutin C, Fayda M, Aksu G, et al. Primary glioblastoma multiforme in younger patients: a single-institution experience. Tumori. 2006;92(5):407–411. doi: 10.1177/030089160609200507. [DOI] [PubMed] [Google Scholar]
- 31.Salmaggi A, Silvani A, Merli R, et al. Multicentre prospective collection of newly diagnosed glioblastoma patients: update on the Lombardia experience. Neurol Sci. 2008;29(2):77–83. doi: 10.1007/s10072-008-0865-x. [DOI] [PubMed] [Google Scholar]
- 32.Hwang SL, Lin CL, Lee KS, et al. Factors influencing seizures in adult patients with supratentorial astrocytic tumors. Acta Neurochir (Wien) 2004;146(6):589–594. doi: 10.1007/s00701-004-0266-8. discussion 594. [DOI] [PubMed] [Google Scholar]
- 33.Walker MD, Alexander E, Jr, Hunt WE, et al. Evaluation of BCNU and/or radiotherapy in the treatment of anaplastic gliomas: a cooperative clinical trial. J Neurosurg. 1978;49(3):333–343. doi: 10.3171/jns.1978.49.3.0333. [DOI] [PubMed] [Google Scholar]
- 34.Lebrun C, Fontaine D, Bourg V, et al. Treatment of newly diagnosed symptomatic pure low-grade oligodendrogliomas with PCV chemotherapy. Eur J Neurol. 2007;14(4):391–398. doi: 10.1111/j.1468-1331.2007.01675.x. [DOI] [PubMed] [Google Scholar]
- 35.Lebrun C, Fontaine D, Ramaioli A, et al. Nice Brain Tumor Study Group Long-term outcome of oligodendrogliomas. Neurology. 2004;62(10):1783–1787. doi: 10.1212/01.wnl.0000125196.88449.89. [DOI] [PubMed] [Google Scholar]
- 36.Grabenbauer GG, Roedel CM, Paulus W, et al. Supratentorial low-grade glioma: results and prognostic factors following postoperative radiotherapy. Strahlenther Onkol. 2000;176(6):259–264. doi: 10.1007/s000660050007. [DOI] [PubMed] [Google Scholar]
- 37.Penfield W, Faulk ME., Jr The insula; further observations on its function. Brain. 1955;78(4):445–470. doi: 10.1093/brain/78.4.445. [DOI] [PubMed] [Google Scholar]
- 38.Isnard J, Guenot M, Ostrowsky K, Sindou M, Mauguiere F. The role of the insular cortex in temporal lobe epilepsy. Ann Neurol. 2000;48(4):614–623. [PubMed] [Google Scholar]
- 39.Isnard J, Guenot M, Sindou M, Mauguiere F. Clinical manifestations of insular lobe seizures: a stereo-electroencephalographic study. Epilepsia. 2004;45(9):1079–1090. doi: 10.1111/j.0013-9580.2004.68903.x. [DOI] [PubMed] [Google Scholar]
- 40.Rossetti AO, Mortati KA, Black PM, Bromfield EB. Simple partial seizures with hemisensory phenomena and dysgeusia: an insular pattern. Epilepsia. 2005;46(4):590–591. doi: 10.1111/j.0013-9580.2005.61904.x. [DOI] [PubMed] [Google Scholar]
- 41.Duffau H. A personal consecutive series of surgically treated 51 cases of insular WHO grade II glioma: advances and limitations. J Neurosurg. 2009;110(4):696–708. doi: 10.3171/2008.8.JNS08741. [DOI] [PubMed] [Google Scholar]
- 42.Moshel YA, Marcus JD, Parker EC, Kelly PJ. Resection of insular gliomas: the importance of lenticulostriate artery position. J Neurosurg. 2008;109(5):825–834. doi: 10.3171/JNS/2008/109/11/0825. [DOI] [PubMed] [Google Scholar]
- 43.Zentner J, Meyer B, Stangl A, Schramm J. Intrinsic tumors of the insula: a prospective surgical study of 30 patients. J Neurosurg. 1996;85(2):263–271. doi: 10.3171/jns.1996.85.2.0263. [DOI] [PubMed] [Google Scholar]
- 44.Neuloh G, Pechstein U, Schramm J. Motor tract monitoring during insular glioma surgery. J Neurosurg. 2007;106(4):582–592. doi: 10.3171/jns.2007.106.4.582. [DOI] [PubMed] [Google Scholar]
- 45.Shankar A, Rajshekhar V. Radiological and clinical outcome following stereotactic biopsy and radiotherapy for low-grade insular astrocytomas. Neurol India. 2003;51(4):503–506. [PubMed] [Google Scholar]
- 46.Crinion J, Ashburner J, Leff A, Brett M, Price C, Friston K. Spatial normalization of lesioned brains: performance evaluation and impact on fMRI analyses. Neuroimage. 2007;37(3):866–875. doi: 10.1016/j.neuroimage.2007.04.065. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.Friston KJ, Holmes A, Poline JB, Price CJ, Frith CD. Detecting activations in PET and fMRI: levels of inference and power. Neuroimage. 1996;4(3, pt 1):223–235. doi: 10.1006/nimg.1996.0074. [DOI] [PubMed] [Google Scholar]
- 48.Hayasaka S, Nichols TE. Validating cluster size inference: random field and permutation methods. Neuroimage. 2003;20(4):2343–2356. doi: 10.1016/j.neuroimage.2003.08.003. [DOI] [PubMed] [Google Scholar]
- 49.Poline JB, Worsley KJ, Evans AC, Friston KJ. Combining spatial extent and peak intensity to test for activations in functional imaging. Neuroimage. 1997;5(2):83–96. doi: 10.1006/nimg.1996.0248. [DOI] [PubMed] [Google Scholar]