Summary: Purpose: Posterior cortex epilepsies (PCEs) encompass a group of epilepsies originating from the occipital, parietal, or occipital border of the temporal lobe, or from any combination of these regions. When their seizures are refractory to pharmacologic treatment, these patients are usually referred for surgery. The aim of our study was to analyze clinical characteristics of all PCE patients referred for surgery from 1994 to 2003, and to search for predictors of surgical outcome.
Methods: We performed a retrospective analysis of clinical and laboratory data from 81 consecutive refractory PCE patients referred for surgery. Surgical and nonsurgical groups of patients were compared, and detailed analyses of all variables of the surgical cases were performed in the search for predictors of seizure outcome.
Results: Risk factors for PCEs included gliosis (34.56%), malformations of cortical development (33.33%), tumors (8.64%), brain trauma (3.70%), Sturge–Weber disease (4.93%), vascular malformations (3.70%), family history of epilepsy (3.70%), history of CNS infections (2.46%), and low IQ (2.46%). Of the 81 patients, 44 were submitted to surgery at the time of the completion of this study. Surgical treatment was highly effective in improving seizures (p < 0.001) when compared with previous pharmacologic treatment alone. Twenty-eight (65.11%) patients became seizure free after surgery versus none in the nonsurgical group. Regarding outcome predictors, patients with shorter duration of epilepsy and those without neurologic abnormalities on clinical examination had higher chances of favorable evolution.
Conclusions: Surgical treatment is effective for the treatment of PCEs and superior to pharmacologic therapy alone. In our series, shorter duration of epilepsy and normal neurologic examination were the only independent variables that predicted better surgical outcome.
Posterior cortex epilepsies (PCEs) encompass a group of epilepsies originating from the occipital, parietal, or occipital border of the temporal lobe, or from any combination of these regions (1). Because no clear anatomic or neurophysiological distinctions are apparent between these cortical areas (2), epilepsies originating from them are probably better analyzed and understood when grouped together.
Patients with pharmacologically resistant PCEs have distinct clinical characteristics and patterns of ictal cortical spread (3–6). Thus the precise diagnosis of PCEs and, moreover, their adequate therapeutic management require an optimized combination of clinical, electrophysiologic, and radiologic studies. However, possibly because PCEs are less common than epilepsies originating from other brain regions, these aspects are less well studied. Although reports of large series can be found in the literature, the impressive and ongoing evolution of diagnostic methods is continuously challenging and changing the way we approach these patients. Therefore reports especially addressing PCE patients whose seizures are refractory to pharmacologic treatment and their surgical outcome are still necessary.
Furthermore, most published series on PCEs refer only to patients who were given surgery, thus limiting their conclusions to a subset of patients possibly with a more focal spectrum of the disease or to obvious surgical cases (for example, tumor). Detailed reports on the whole spectrum of refractory PCE patients, including more complex cases who also had extensive presurgical evaluation but whose surgery was delayed or deferred, are clearly lacking. Here we describe a series of such patients with refractory PCE, including surgical and nonsurgical cases, and also searching for preoperative variables that best predict postoperative seizure outcome.
We retrospectively analyzed presurgical data of 1,537 patients evaluated at the Ribeirão Preto Epilepsy Surgery Program (São Paulo, Brazil), from July 1994 to May 2003. Patients that fulfilled at least two of the following inclusion criteria were selected for the present study: (a) seizure semiology consistent with PCEs: patients with visual or somatosensory aura; (b) ictal or interictal EEG consistent with posterior cortex involvement; and (c) posterior cortex magnetic resonance imaging (MRI) lesions. We excluded patients whose clinical and laboratory data raised doubts on the diagnosis of PCEs, as well as cases with additional epileptogenic lesions outside the posterior cortex.
Of the 1,537 patients, 81 (5.27%) patients fulfilled our inclusion criteria and were selected for this study (Table 1). Patients were evaluated according to previously reported standardized protocols approved by the Ethics Committee of our institution. At the time of the completion of this study, 44 (54.32%) patients had been given surgery, whereas 37 (45.68%) patients had not (nonsurgical group).
Table 1. Demographic data
|Family history of epilepsy: Yes/No||28/53|
|Mean age at epilepsy onset (yr) (SD)||7.02 (7.92)|
|Mean duration of epilepsy (yr) (SD)||14.70 (11.54)|
|Mean age at evaluation (yr) (SD)||21.71 (13.11)|
|Seizure frequency: Daily/Other||43/38|
Presurgical evaluation included a detailed clinical history and neurologic examination, interictal scalp EEG recording, video-EEG monitoring (video-EEG), structural and functional imaging, neuropsychological testing, and, whenever needed, intracarotid amobarbital testing (Wada test) for memory and speech representation. Neuroimaging studies included 1.5-T high-resolution MRI with dedicated protocols for epilepsy and ictal and interictal single-photon emission computed tomography (SPECT) scans. In the surgical group, 34 (77.27%) patients had intraoperative electrocorticography (ECoG), and 13 (29.54%) patients were additionally given invasive video-EEG monitoring with subdural electrodes before definitive surgery.
Seizures were initially analyzed through evaluation of clinical history and medical records. Detailed video-EEG analyses of 783 events were independently performed by two board-certified electroencephalographers (T.R.V., C.L.D.), according to a standardized protocol. Initial ictal symptoms (e.g., visual aura, somatosensory aura) or signs (tonic, tonic–clonic, versive, clonic) or both were evaluated according to the standards of the International League Against Epilepsy (7,8).
High-resolution MRI (1.5-T Siemens Magneton Vision; Siemens, Munich/Erlangen, Germany) were independently interpreted by two neuroradiologists with experience in epilepsy neuroimaging (A.C.S., D.A.). MRI lesions were classified into the following subgroups: parietal, occipital, parietooccipital, temporooccipital, and temporoparietooccipital regions. Both pre- and postsurgical MRIs were compared to determine whether the resections were complete or incomplete.
Video-EEG duration ranged from 3 to 15 days (mean, 5 days). EEG signals were digitally recorded, and synchronized video was acquired on an S-VHS Panasonic recorder (Vangard Systems, Cleveland Clinic Foundation, Cleveland, OH, U.S.A.). Scalp electrodes were placed according to the 10–20 system plus additional close-spaced electrodes over the posterior regions. In total, 783 events similar to patients' habitual seizures were recorded (mean, nine seizures per patient; range, one to 49 seizures).
Interictal spikes were independently reviewed by two board-certified electroencephalographers (T.R.V., V.C.T.). Their frequency, ratio, and localization were visually assessed. Five-minute EEG samples were analyzed every hour, 24 h/day, totaling 120 min/day. Spikes and sharp waves were defined according to the guidelines of the International Federation of Clinical Neurophysiology (9).
Ictal EEG patterns were classified as follows: (a) focal, if the ictal activity was maximum at a single electrode and no more than two contiguous electrodes were within 80 to 100% of the maximal amplitude; (b) regional, when the ictal activity involved electrodes overlying a single lobe and had ≥2:1 amplitude predominance in electrodes placed over other regions of the same hemisphere; (c) posterior quadrant, when the ictal activity involved multiple electrodes over the posterior cortex; and (d) diffuse, when the ictal activity involved multiple electrodes over both cerebral hemispheres. The epilepsy-onset zone was independently assessed on ictal video-EEG by the previously mentioned investigators. Whenever the results were discordant, they were reviewed by the two examiners together to reach a consensus. Ictal propagation was not evaluated in this study.
Specimens of all surgically treated patients were analyzed for detailed histologic study. The findings were classified as gliosis (caused by previous encephalitis, trauma, stroke, previous surgical resection, and others), vascular malformation, malformation of cortical development (MCD), Sturge–Weber disease, and brain tumors (dysembryoplastic neuroepithelial tumors or low-grade gliomas).
Risk factors for pharmacologically resistant PCEs
All risk factors for epilepsy were annotated according to Hauser (1999) (10). Because no single factor could be assigned as a sine qua non condition for refractory PCE, we avoided the use of the term “cause of epilepsy”. All patients were reviewed, with special emphasis on the following risk factors: CNS infection, stroke, brain trauma, brain tumor, degenerative disease, low IQ (<70) MCD, gliosis, family history of epilepsy, obstetric complications, and febrile seizures. When two risk factors were identified, the one judged more related to the pathology was considered. For example, if a patient had both febrile seizures and MCD, we considered MCD to be the risk factor.
Surgery and surgical outcome
All surgical complications, major or minor, were reviewed to establish the morbidity of PCEs surgery. Patients were additionally classified into two subgroups: seizure free (Engel class I) and not seizure free (Engel classes II, III and IV) (11). Postsurgical isolated auras were not considered for this outcome purpose. All surgical patients had ≥1 year of follow-up (mean follow-up time, 39.7 months).
Categorical data of patients were compared through the two-tailed Fisher's exact test. Mean numerical variables were compared by the Student independent t test. We additionally examined the independent effect of the variables as possible predictors of surgical outcome through binary logistic regression analysis by using the statistical program SPSS version 10.0 (Chicago, IL, U.S.A.). The magnitude of the association between selected variables and surgical outcome was measured by the crude and adjusted odds ratio and respective 95% confidence intervals. To determine the number of independent variables to be included in our logistic regression model, we used the parameters suggested by Stevens (12). Results were considered significant if p < 0.05. Bonferroni's adjustments for p value were not applied because they are not appropriate for biomedical research and lead to unacceptable increase of type II error (13,14).
Of 1,537 patients evaluated at the Ribeirão Preto Epilepsy Surgery Program (São Paulo, Brazil) from 1994 to 2003, 81 (5.27%) patients, 46 male and 35 female patients, fulfilled predefined criteria for PCE.
Gender, family history of epilepsy, daily seizures, age at epilepsy onset (mean, 7.02 years; range, first year of life to 40 years), epilepsy duration (mean, 14.70 years,;range, 16 months to 44 years), and age at presurgical evaluation (mean, 21.71 years; range, first year of life to 52 years) are presented in Table 1. Forty-four patients were submitted to surgery at the time of the completion of this study. One patient died during the immediate postoperative period, thus long-term follow-up was available for 43 patients. The remaining patients were not submitted to surgery for reasons explained further.
Regarding the surgical outcome (Table 2), 28 (65.11%) patients of 43 surgically treated became free of seizures (p < 0.001, when compared with the previous status of these patients). The side of the lesion, age at seizure onset, age at surgery, presence or absence of auras, presence of ictal clinical lateralizing signs, low IQ, and comorbid psychiatric symptoms were not associated with surgical outcome (see Table 2). Although not significant, longer duration of epilepsy and gliosis on neuropathologic examination showed a statistical trend for unfavorable surgical outcome (p = 0.07 and p = 0.07, respectively). Abnormalities on neurologic examination and higher seizure frequency (daily vs. nondaily seizures) showed a statistical significance for poor surgical outcome (p = 0.02 and p = 0.02, respectively), but after logistic regression (Table 3), only duration of epilepsy and neurologic status during presurgical evaluation remained as significant predictors of surgical outcome, with patients with shorter duration of epilepsy or those with normal neurologic examination having better chances of reaching seizure-free status after surgery (p = 0.049 and p = 0.02, respectively).
Table 2. Surgical outcome
|Number of patients||43||28||15||-|
|Family history of epilepsy: Yes/No||43||12/16||7/8||1.00|
|Mean age at seizure onset (yr; SD)||43||7.7 (±8.9)||7.3 (±9.7)||0.87|
|Mean age at surgery (yr; SD)||43||20.3 (±10.9)||26.2 (±14.4)||0.13|
|Duration of epilepsy (yr; SD)||43||10.9 (±8.7) ||18.0 (±12.9)||0.07c|
|Clinical lateralizing: Present/Absent||43||16/12||9/6||1.00|
|Neurologic exam: Normal/Abnormal||43||22/06||6/9||0.02 c|
|Low IQ: No/Yes||43||21/07||8/7||0.18|
|Associated psychiatric symptoms: No/Yes||43||21/07||11/04||1.00|
|Seizure frequency: Daily/Nondaily||43||14/14||13/02||0.02 c|
|EEG baseline activity: Normal/Abnormal||43||10/18||2/13||0.16|
|Interictal EEG: Localizing/Nonlocalizinga||42||15/13||8/6||1.00|
|Ictal EEG: Localizing/Nonlocalizinga||42||19/9||7/7||0.32|
|MRI lesion localization:|
|MRI lesion localization: Infrasylvian/Other||43||8/20||1/14||0.12|
|MRI control: Complete/Incomplete||43||12/16||4/11||0.34|
|Type of surgery|
| Limited resection/Large resection||43||17/11||8/7||0.75|
| Gliosis/Other||43||4/24||6/9||0.07 c|
| Vascular malformation/Other||43||2/26||0/15||0.53|
| Sturge–Weber disease/Other||43||3/25||1/14||1.00|
Table 3. Logistic regression of surgical outcome predictors
|Abnormal neurologic exam||5.50||(1.40–21.68)||0.02||12.52||(1.57–87.10)||0.02a|
|Duration of epilepsy||0.93||(0.88–1.00)||0.07||1.10||(1.00–1.21)||0.049a|
|Gliosis at neuropathology||4.00||(0.91–17.55)||0.07||2.46||(0.31–19.25)||0.39|
Table 4 summarizes the characteristics of the auras and seizure phenomena presented by our patients. Visual auras occurred in 23 (28.39%) patients, somatosensory auras in 12 (14.81%), and other types of auras in 17 (20.99%) patients. No aura was reported in 29 (35.80%) patients. No particular subtype of aura was indicative of any particular cortical area.
Table 4. Localization of MRI lesion versus aura and seizure pattern
|Parietal (17)||Visual hallucination, 4||Tonic, 6|
|Visual loss, 1||Clonic, 6|
|Somatosensory, 7||Complex partial, 5|
|Dizziness, 1|| |
|Cephalic, 1|| |
|Movement sensation, 1|| |
|No aura, 2|| |
|Occipital (15)||Visual hallucination, 5||Tonic, 3|
|Visual illusion, 2||Clonic, 8|
|Visual loss, 2||Versive, 2|
|Dizziness, 1||Complex partial, 2|
|Movement sensation, 1|| |
|No auras, 4|| |
|Parietooccipital (08)||Visual hallucination, 2||Tonic, 3|
|Dizziness, 1||Clonic, 2|
|Somatosensory, 2||Versive, 1|
|No auras, 3||Complex partial, 2|
|Temporoparietooccipital (25)||Visual hallucination, 2||Tonic, 12|
|Somatosensory, 1||Clonic, 5|
|Visual loss, 1||Versive, 2|
|Dizziness, 1||Complex partial, 6|
|Blinking, 1|| |
|Movement sensation, 3|| |
|Not specific, 1|| |
|No auras, 15|| |
|Temporooccipital (09)||Visual hallucination, 2||Clonic, 2|
|Visual illusion, 1||Complex partial, 7|
|Visual loss, 1|| |
|Movement sensation, 2|| |
|No auras, 3|| |
|No lesion (07)||Somatosensory, 2||Tonic, 2|
|Blinking, 1||Clonic, 1|
|Dizziness, 1||Complex partial, 4|
|Movement sensation, 1|| |
|No auras, 2|| |
Regarding seizure phenomena, 26 (32.10%) patients had complex partial seizures, 26 (32.10%), tonic seizures, 24 (29.63%), clonic seizures, and five (6.17%), versive seizures. When the lesion was located in the temporooccipital region, no tonic seizure was observed. Indeed, when lesions were localized below an imaginary line drawn at the level of sylvian fissure, patients had significantly fewer motor phenomena (p < 0.01). No characteristic of the seizures observed after the aura was indicative of any particular cortical area. Tonic seizures showed a statistical trend for poorer outcome. All other seizure patterns did not show any correlation with outcome.
EEG findings are shown in Table 5. Interictal abnormalities were observed in 73 (91.25%) of 80 patients given video-EEG monitoring. Distribution of interictal abnormalities (whether focal, regional, posterior quadrant, or diffuse) showed that in only a small number of patients were the interictal abnormalities focal (7.5% of the patients). Localizing EEG (focal plus regional plus posterior quadrant) was observed in 49 (61.25%) patients. Interictal EEG showed no postsurgical prognostic value (see Table 2).
Table 5. Electroencephalographic findings
|Normal||7 (8.8%)||6 (7.5%)|
|Focal||6 (7.5%)||4 (5.0%)|
|Regional||19 (23.8%)||19 (23.8%)|
|Posterior quadrant||24 (30.0%)||28 (30.0%)|
|Diffuse||24 (30.0%)||23 (28.8%)|
|Total||80 ||80 |
Ictal EEGs were abnormal in 74 (92.5%) patients. Distribution of ictal abnormalities (whether focal, regional, posterior quadrant, or diffuse) showed that in only a small number of patients were the abnormalities focal (5% of the patients). Localizing ictal EEG (focal plus regional plus posterior quadrant) was observed in 51 (63.75%) patients. As observed for interictal, ictal EEG was of no postsurgical prognostic value (see Table 2).
Risk factors for PCEs
Risk factors for PCEs included gliosis (34.56%), MCDs (33.33%), tumors (8.64%), Sturge–Weber disease (4.93%), brain trauma (3.70%), vascular malformations (3.70%), family history of epilepsy (3.70%), central nervous system infections (2.46%), and low IQ (2.46%) (Table 6). Patients with gliosis were significantly less frequent in the surgical group, whereas patients with MCDs were significantly more represented in the surgical group (p = 0.003).
Table 6. Risk factors
|Malformation of cortical development||07||20||27|
| Brain trauma||02||01||03|
| Sturge–Weber|| 0||04||04|
| Vascular malformation||01||02||03|
| CNS infection/Other||01||01||02|
|Family history of epilepsy: Yes/No||03|| 0||03|
|Low IQ: Yes/No||02|| 0||02|
|No risk factor: Yes/No||02|| 0||02|
All patients given surgery had lesions disclosed by MRI. Histologic analysis (see Table 6) of resected tissue confirmed MCDs in 20 patients (45.45% of surgical cases) as the most common etiology in our patients. Other pathologies were gliosis in nine patients (20.45% of surgical cases), tumors in six (13.63% of surgical cases), Sturge–Weber disease in four (9.09% of surgical patients), and vascular malformations in two patients (4.54% of surgical cases). Within the tumors, we encountered four dysembryoplastic neuroepithelial tumors and two low-grade astrocytomas.
Types of resections
Postoperative (PO) MRIs were obtained for all but one patient who died on PO day 21. Multilobar resection was performed in 19 (44.18%) of 43 patients, lesionectomy in 17 (39.53%), and lobectomy in five (11.63%). According to PO MRI, surgeries were considered complete in 16 (37.21%) and incomplete in 27 (62.79%) of 43 patients. However, the completeness of the resection had no influence on surgical outcome (p = 0.34; see Table 2).
Thirty-seven patients were not given surgery until the end of this study. The main reason for delaying surgical treatment in these patients was the need for invasive electrodes in 28 (75.68%) patients. In the remaining patients, surgery could not be performed at the time of this survey for several other reasons. One patient improved seizure control while on the ketogenic diet. One surgery was not performed because of unacceptable risk of major visual deficit. One patient had a low-grade and nonprogressive tumor in the vicinity of eloquent cortex (Wernicke area), and for this reason was not given surgery at the time of completion of this study. Finally, six patients refused any surgical procedure including the implantation of intracranial electrodes.
Complications of the surgical procedure
Absence of any type of PO complication, including increase of visual deficits, was observed in 20 (45.45%) patients. Visual field deficits were already detected in eight (18.18%) patients at their preoperative evaluation. In 11 (25.00%) patients, de novo visual field defects developed; three (6.82%) had postsurgical subgaleal hematoma that required surgical drainage; two (4.54%) had meningitis after surgery; in two (4.54%), acute subdural hematomas developed; one (2.27%) had epidural empyema plus scar infection; one (2.27%) had CSF fistulae; in one (2.27%), deep venous thrombosis developed; one (2.27%) had pneumoencephalus, and one (2.27%) had status epilepticus after surgery. Two (4.54%) patients needed intraoperative blood transfusion, and one (2.27%) patient had brainstem vascular ischemia and unfortunately died on PO day 21. Two patients had more than one complication. Except for new-onset visual field defects and one death, all other patients had complete resolution of their PO complications.
We present a large series of consecutive patients with pharmacologically intractable PCEs who were referred for surgical treatment at our center. In our study, we analyzed surgical as well as nonsurgical patients, at the time of this survey. This approach was chosen to avoid the spectrum bias that happens when investigators include only “clear-cut” or “well-defined” cases that are not representative of the whole spectrum of the disease (15). Overall, 28 patients from 44 surgical cases became seizure free. The remaining 15 patients did not become seizure free but, rewardingly, all of them experienced improvement in their seizure frequency. Although the two groups of patients were not randomized, surgical treatment of PCE emerged as highly effective in improving seizures not only in intergroup comparison (surgical vs. nonsurgical) but also in intragroup comparison (surgical group, preoperative pharmacologic treatment alone vs. surgery).
One of the relevant findings was the observation that the etiologic profile of pharmacologically intractable PCEs in a developing country is similar to the one observed in industrialized countries (16,17). Because our center is a national referral center for epilepsy surgery in the country and attracts patients from all its regions, it is reasonable to assume that our series truly represents pharmacologically refractory PCEs in the Brazilian population and does not reflect any particular subgroup of patients.
The observation that patients from developing and developed countries have the same etiologies of intractable PCEs is an interesting feature. We would expect that infectious diseases traditionally associated with epilepsy, such as cysticercosis, tuberculosis, echinococcosis, and other tropical diseases, would be more prevalent in our series than in those from developed countries (16,17). However, our data did not confirm this expectation. One possible explanation is that these infectious etiologies may well be risk factors only for epilepsy but not for medical intractability. In pathologically confirmed cases, MCDs were observed in 45.45% of the surgical patients, gliosis in 20.45%, tumors in 13.63%, Sturge–Weber disease in 9.09%, and vascular malformations in 4.54%. Absence of any identifiable causes was observed in only two (2.46%) of all patients. In other series, tumor cases were encountered in 22–37% and developmental abnormalities in 9–40% of the patients (1,4,18). In a recent report published by Boesebeck et al., (17), the main etiologies of refractory PCEs were gliosis (40%), tumors (23%), MCDs (23%), and vascular lesions (12%).
It also is interesting to observe that when lesions were localized below an imaginary line drawn at the level of the sylvian fissure, patients had fewer motor phenomena. Seizures originating from this inferior region seemed to have better chances to propagate to the temporal lobe, whereas seizures emanating from other posterior cortex areas had probably more complex and diffuse propagation patterns, involving motor areas more easily, in agreement with previous findings of Ajmone-Marsan and Ralston (19).
Regarding surgical outcome, patients with no abnormalities on routine neurologic examination or those with less frequent seizures seemed to have significantly better surgical outcome, whereas patients with gliosis and longer duration of epilepsy had only a trend toward unfavorable outcome. However, after logistic regression, only those patients with shorter duration of epilepsy or normal neurologic status were more likely to become seizure free after surgery. The fact that abnormal neurologic status correlates with poor surgical outcome suggests that lesions producing neurologic deficits are more difficult to excise completely, possibly because of their extension and involvement of eloquent cortex.
In our series, as also observed by others, ictal semiology was not a predictive factor of seizure-free status (17). Other presurgical variables like gender, family history of epilepsy, side of the lesion, presence or absence of auras, lateralizing signs based on clinical findings, low IQ, EEG findings, localization of the lesion within the posterior cortex, completeness of the resection, and large versus smaller resections also were not associated with surgical outcome. Scalp EEG data agreed with those reported by some authors (17) but disagreed from others (1). Contrary to recent findings of Barba et al. (20) in a retrospective study involving 14 patients, we also did not find significant correlation between completeness of MRI lesion resection and surgical outcome, in line with previous observations that the remaining epileptogenicity may arise from cortical areas that are not part of the MRI-defined lesions (17). However, it is possible that our study lacks statistical power for defining this question, and larger series or even future meta-analyses may well be necessary for clarifying this question further. In contrast to our findings, some authors reported that surgical outcome may be correlated with some specific type of pathologies, such as MCDs, that have good outcome rates ranging from 40 to 49% (16,21,22), whereas tumor ranges from 77 to 80% (4,17,18,21). Although our rates for good outcome were similar to those reported in the literature, with 80% for tumors, 67% for MCDs, 100% for vascular malformations, and 75% for Sturge–Weber disease, no particular type of pathology could be statistically correlated with surgical outcome. This fact may be explained by the size of our sample. Future meta-analysis might possibly clarify this matter as well.
Our results regarding history data showed that, except for patients with longer epilepsy duration or high seizure frequency who seemed to have a worse prognosis, no other variables were correlated with surgical outcome, in line with Boesebeck et al. (17). In relation to the age at epilepsy onset that Blume et al. (1) reported might be of prognostic value, we did not observe this correlation, but noticed that duration of epilepsy can be related to surgical prognosis in PCEs. If confirmed by other studies, this finding might constitute a strong argument for early surgical intervention in these patients.
Few data address surgical complications of PCE surgery, and more reports are clearly needed. When described, complications mainly refer to visual field deficits rather than other surgical complications. In our series, 54.55% of the patients had some type of comorbidity associated with surgery, if all surgical complications, permanent and nonpermanent, minor and major complications were considered. Permanent complications, however, occurred much less frequently; 25% of the patients had de novo visual field defects, and one patient died. These observations also are in line with other reports. For example, Olivier and Bolin (23) mentioned that 24% of the patients experienced some comorbidity associated with surgery for treatment of parietal lobe epilepsy. Thus it seems that surgical complications in PCE epilepsy are indeed elevated, and this aspect should be carefully considered when this type of surgery is planned.
In conclusion, important observations came from this study. First, our data provided further evidence that surgical treatment of PCEs is efficient in controlling seizures when compared with modern pharmacologic treatment alone. Second, those patients with early surgical interventions and those without abnormal findings in neurologic examinations have better outcomes. Third, and contrary to previous expectations, the etiologic profiles of PCE epilepsy in developing and developed countries are similar. Fourth, PCE surgery carries higher risks of permanent and nonpermanent surgical complications. In spite of our conclusions, we recognize that, as in other studies, the relatively small samples of PCE series preclude some other conclusions that might be possible. Thus larger series or meta-analyses from several independent centers are still needed the better to delineate the real picture of refractory epilepsy of the posterior cortex and to help in defining better treatment strategies for each particular etiology or subregion.
Acknowledgment: CNPq, Brazil (Drs. Dalmagro, Walz, and Bianchin); FAPESP, Brazil (Dr. Bianchin 02/03743-0); CAPES, Brazil (Dr Wichert-Ana); and FAEPA, Brazil, supported this study. We thank Mrs. Electra Greene for her professional review of the English version.