Precise outcome data about the surgical therapy of cerebral cavernous malformation (CCM)–associated epilepsy is scarce regarding different epilepsy types, surgical approach, and outcome. Long-term outcome in patients with CCM-associated epilepsy is analyzed in a large single-center series.
Seizure outcome data >24 months was available in 118 patients. The influence of different parameters of preoperative workup and surgical technique was analyzed with regard to seizure outcome.
The study cohort comprised 76 patients with drug-resistant epilepsy (DRE), 20 patients with chronic epilepsy that did not meet the definition of DRE, as well as 22 patients with sporadic seizures. Temporal localization of CCMs predisposed to develop DRE. Detailed epileptologic workup was performed in 85 patients; invasive monitoring was done in 23 (37%) of 76 DRE cases. In 84% of DRE cases more extensive resections were performed. Mean follow-up varied between 107 and 137 months for the three groups. Seizure freedom in DRE was 88%, in chronic epilepsy 80%, and in sporadic seizures was 91%. Longer symptom duration was associated with worse seizure outcome. Outcome of patients who underwent invasive monitoring was not worse. The outcome in CCM-associated DRE can be good if more extensive resections are used and if noninvasive and/or invasive presurgical epileptologic workup is used whenever indicated. DRE was considerably more frequent in the temporal lobe, suggesting that temporal localization predisposes development of DRE. Seizure freedom rates were stable over a long period.
Surgical therapy of CCM-associated seizures and epilepsy can be successful if different surgical techniques according to presurgical evaluation are realized. To prevent clinical worsening into DRE, surgical intervention in CCM-associated epilepsy may be considered early.
Vascular malformations in the adult brain are often accompanied by epileptic seizures and represent a common neurosurgical problem. Cerebral cavernous malformations (CCMs) account for 10–15% of all vascular malformations in the adult brain (Batra et al., 2009). The incidence of CCMs is thought to range between 0.4% and 0.8% (Del Curling et al., 1991). Forty percent to 70% of patients who have supratentorial CCM present with epilepsy (Awad & Jabbour, 2006; Ferrier et al., 2007). The risk for seizure development is estimated between 1.5% and 2.4% per patient and year (Del Curling et al., 1991; Moriarity et al., 1999). CCMs do not have any intrinsic epileptogenicity. Seizures most probably result from various effects of blood breakdown products in the perilesional cortical area (Kraemer & Awad, 1994; Raabe et al., 2012). The epileptogenicity of CCM is influenced by its localization, particularly archicortical or temporal lobe localization (Menzler et al., 2010).
Patients with symptomatic CCM may present clinically with occasional epileptic seizures but they may also lead to chronic or even drug-resistant epilepsy (DRE) in about 35–40% of all cases (Kondziolka et al., 1995; Porter et al., 1997; Chang et al., 2009). Up to 4% of all DRE patients were diagnosed with a CCM (Kuzniecky et al., 1987; Convers et al., 1990). Singularity of CCM as well as medically controlled seizures may be positive predictors of seizures outcome (Englot et al., 2011).
There are two main causes to consider resection of symptomatic CCM: to prevent renewed symptomatic hemorrhage, which can cause persistent neurologic deficits, and to cure structural DRE. In some cases mixed recommendations were given by neurosurgeons and epileptologists weighting between epilepsy without final approval of drug resistance against risk of renewed bleeding.
Seizure outcome after epilepsy surgery can be favorable in patients with single supratentorial CCMs compared with conservative treatment with antiepileptic drugs (AEDs) or other treatment modalities like irradiation (Robinson et al., 1991; Shih & Pan, 2005; Hsu et al., 2007). Microsurgical removal of CCM is a well-established treatment both for cases with sporadic seizures and for chronic and/or DRE. Up to date there is still debate if the surgical removal of the hemosiderotic rim around the CCM is making seizure outcome more favorable. Many case series of CCM surgery have been published, but the current literature often does not include sufficient description of the surgical method used, the considerations defining the area of resection, and detailed information on extent of resection as recently described in an analysis of CCM series (von der Brelie & Schramm, 2011). Published overall seizure outcome rates are often not useful, since authors most probably mix their patient populations regarding the type of epilepsy. Larger case series are needed that distinguish clearly between DRE and other, milder forms of epilepsy (occasional seizures or treatable but persistent epilepsy).
Cases of DRE require epilepsy surgery, which normally includes carefully presurgical epileptologic evaluation. One may argue if there is need for extended presurgical epileptogenic workup or whether it would be feasible to perform surgery without that.
Are there differences in seizure outcome between patients with and without formal presurgical evaluations? Are there differences in seizure outcome regarding the presurgical epilepsy type? Is the localization of CCM predictive for seizure type and postsurgical seizure outcome? The aim of this study is to answer these three relevant questions in a large patient cohort from the Bonn epilepsy and cranial surgery database.
The Bonn neurosurgical index database and the Bonn epilepsy surgery database were searched from 1988 to 2010. A total of 168 patients with CCM-associated epilepsy were identified. All biopsy specimens underwent neuropathologic analysis, and only cases with concordant neuropathologic, neuroradiologic, and intraoperative findings, particularly unequivocal exclusion of arteriovenous malformation, were considered for this study.
From the 168 overall patients with CCM-associated epilepsy, 20 patients were <19 years old and were thus excluded from the analysis. Fifteen patients harbored multiple CCMs and were thus excluded. Fifteen patients with a follow-up duration <24 months were excluded. Patients with deep-seated CCMs located in basal ganglia, thalamus, cerebellum, purely white matter, brainstem, or spinal CCM were explicitly not analyzed. In the end, the study population consisted of 118 patients.
Analyzed data included patient age, sex, age at onset of seizures, and duration of symptoms. Localization of CCM was analyzed based on radiologic images, radiologic reports, and surgical records. Temporal localization was subdivided into temporolateral neocortical and temporomesial archicortical localization. Localization was categorized as to be in eloquent if CCM was located the precentral gyrus, in the postcentral gyrus, or in Wernicke's or Broca's area of the dominant hemisphere, as well as in the primary visual cortex.
Age at surgery, symptomatic hemorrhage, presurgical epileptologic workup, surgical technique and extent of resection, surgical morbidity and mortality, as well as seizure outcome was analyzed.
Designation of type of epilepsy
Drug-resistant epilepsy (DRE) was defined as having seizures for a minimum duration of 2 years despite adequate treatment with at least two different, appropriate AEDs. Patients with persisting seizures without matching the criteria of DRE were categorized as having chronic epilepsy. Patients were designated as having sporadic seizures if they had one or two seizures. The latter categorization does not correspond to the classical epileptologic definition of epilepsy types. We chose this operational categorization on purpose based on the three most frequent patient groups coming to a neurosurgeon for counseling: DRE cases, chronic epilepsy, and patients with one or two seizures (sporadic seizures).
Presurgical epileptologic workup
Patients with DRE underwent detailed epileptologic evaluation that assessed seizure semiology, magnetic resonance imaging (MRI) findings, and video–electroencephalography (EEG) monitoring, and sometimes invasive monitoring. Invasive monitoring was done exclusively in patients with DRE, and it was carried out in two constellations:
if clinical or imaging findings and electrophysiologic monitoring revealed incongruent results
if the lesion was located in or close to eloquent brain areas so that functional mapping should distinguish between lesional epileptogenic zone and functional area
It might also be that these two aspects have been addressed simultaneously in the same patient. Standards and techniques of presurgical evaluation have been published (Kral et al., 2002).
Resection types were divided into pure lesionectomy (hemosiderotic rim was not or not completely removed), lesionectomy including rim, and extended resections. Extended resections were subdivided into tailored resections (lesionectomy including rim and adjacent cortex), lesionectomy including standard mesial resection (cases in which the lesion and the hemosiderotic rim as well as amygdala and hippocampus/parahippocampus were also removed), and standard temporal lobe resections (including classical one-third or two-third resections). Because all senior neurosurgeons in this unit were familiar with the principles of epilepsy surgery it was a principle to remove the hemosiderotic rim unless the hemosiderotic rim was involving eloquent areas. The rim was always removed completely, even in the depth of the sulcus. Most cases have been operated by epilepsy surgeons anyway. Therefore, this principle was followed strictly and documented in the surgical records. T2* imaging to quantify hemosiderotic remnants postoperatively was not done in most of the patients included in this series. Type of resection was evaluated by analysis of surgical records and in some cases by postoperative MRI.
Follow-up and seizure outcome
Follow-up was obtained by seeing the patients at 3 months and 1 year after surgery. In addition, all patients, their families and/or their general practitioners/neurologists were contacted by phone and asked to provide information on further seizure history and antiepileptic medication utilizing a standardized interview. Using this strategy a year-to-year follow-up was generated retrospectively. Seizure outcome was determined using the Engel and the International League Against Epilepsy (ILAE) classification (Engel et al., 1987; Wieser et al., 2001). Double losers were defined as patients who developed a new and permanent postoperative deficit and who were not seizure free after surgery.
Descriptive statistics (means, medians, and percentages) were obtained. Testing of significance was assessed at p < 0.05 using chi-square/Fisher's exact test, Mann-Whitney U-test, and Student's t-test (SPSS version 20; IBM, Somers, NY, U.S.A.).
The study population comprises a total of 118 patients (see 'Methods'). Of this cohort, 71 patients (60%) were male. Mean age at surgery was 38.9 years (range 20–72 years, standard deviation [SD] 12.2 years (see Table 1).
Table 1. Demographic data
Mean age/onset of seizures (years)
Mean age/surgery (years)
Mean duration of symptoms (years)
Epilepsy type/seizure semiology
Seventy-six (76/118 = 64%) patients were classified as having DRE, 20 (20/118 = 17%) patients had chronic epilepsy, and 22 (22/118 = 19%) patients had sporadic seizures (see Table 1). Patients with DRE had a lower mean age of seizure onset (t-test, p=0.001). Symptom duration was longer in patients with DRE (t-test, p = 0.01).
Sixty-six (87%) of DRE patients had complex partial seizures; 10 patients (13%) had simple partial (e.g., motor or visual) seizures. Fifty-nine (78%) of these patients had also generalized convulsions, 87.5% of which were secondarily generalizing (see Table 2 for other seizure groups).
Table 2. Epilepsy/seizure semiology
Chronic epilepsy (%)
Sporadic seizures (%)
TBI, traumatic brain injury.
Complex partial seizures
Other focal seizures
Secondary generalizing convulsions
Mesial temporal sclerosis
Other pathologies (tumor, vascular, TBI)
In the DRE group, six patients (8%) did also display mesial temporal sclerosis representing a potential secondary focus of epileptogenesis in dual pathology.
Localization of the symptomatic CCM
In the overall population, 57% had a CCM in the temporal lobe (see Fig. 1). Regarding temporally located CCMs, 52% were located in the lateral temporal lobe and 48% in the mesial temporal lobe structures, respectively.
Fifty-five patients (72%) with DRE had a temporal lobe CCM. In 13 patients of the DRE group (17%), the CCM were located in eloquent brain structures and 8 patients (11%) harbored a CCM in noneloquent and nontemporal brain area.
In contrast to this group of patients with DRE, a different distribution of symptomatic CCMs was found in patients with chronic and sporadic seizure; seven patients (35%) with chronic epilepsy had a temporal CCM. In the sporadic seizure group, eight patients (36%) had a temporal CCM (see Fig. 1). Temporal localization was much more frequent in DRE. Patients with temporal lobe CCMs are prone to develop DRE (p = 0.01).
Detailed epileptologic workup in CCM cases must establish the epileptogenic zone and its relationship to the CCM. In 12 patients (16%) with DRE, electrophysiologic diagnostics showed lack of concordance between localization of the CCM and ictal focus. In these patients, invasive monitoring was performed (see next section).
Detailed epileptologic workup was done in four patients with chronic epilepsy (20%) and in one patient with sporadic seizures (5%), respectively.
In 23 (28%) of 76 DRE cases, invasive EEG monitoring was performed. In 40% of these cases functional grid mapping of eloquent brain areas was performed simultaneously. None of the patients with chronic epilepsy or sporadic seizures was implanted.
Statistically, invasive monitoring did not improve postoperative seizure outcome in the groups with more longstanding types of epilepsy.
Microsurgery was used in all cases. Most of DRE cases (66 cases; 87%) were operated performing extended resections. Classical one third or two third anterior temporal lobectomy including rim and lesion was performed in 15 (26%) of 61 operations in the temporal lobe. In 7 (19%) of 37 patients with neocortical temporal CCMs, mesial structures were also removed because presurgical epileptologic evaluation suggested their involvement. In nine cases (12%), multiple subpial transsections (MSTs) in adjoining functional cortex were also carried out. In DRE cases only one patient underwent pure lesionectomy.
Eleven patients with DRE (14%) received lesionectomy including rim. Intraoperative electrocorticography (ECoG) was performed in three patients with DRE (4%). Neuronavigation technique was used in 21% (see Fig. 2).
In the other groups of epilepsy, operative strategy was different. In the chronic epilepsy group (n = 20), six patients (30%) underwent extended resections. Eleven patients (55%) underwent lesionectomy with rim and the remaining 15% underwent pure lesionectomy (see Fig. 2).
Most patients (20 of 22 patients, 91%) with sporadic seizures underwent lesionectomy including the hemosiderotic rim (see Fig. 2). In summary, the more longstanding the clinical symptoms of epilepsy, the more frequent are more extensive resections.
Intraoperative electrophysiologic evaluation motor evoked potentials/sensory evoked potentials (MEP/SEP) was used in 11 patients (9%) to delineate functional motor cortex or monitor motor function.
Regarding the overall study population, 29 patients (24.6%) developed a postoperative neurologic deficit, 8 of which (7% of all patients) developed a permanent postoperative neurologic deficit (one homonymous heminanopia, three motor deficits, three sensory deficits, one speech deficit), excluding accepted visual field defects. Twenty-one patients developed an accepted upper or lower quadrant anopia (18%).
In 54 of 76 surgical procedures for DRE the postoperative course was uneventful. In 22 cases (29%) patients developed a neurologic deficit, which consisted in 12 in upper or lower quadrantanopia as a calculated (i.e., accepted) deficit. One patient had complete hemianopia. Ten patients developed motor or speech deficits, of which 90% were transient. Therefore, 2 (3%) of 76 patients in the DRE group had permanent noncalculated deficit.
Twenty percent of all patients with chronic epilepsy developed a permanent, noncalculated deficit. The rate of noncalculated neurologic deficit in the sporadic seizure group was 9%.
Postoperative seizure outcome
Overall, in the whole study population ILAE 1a outcome (patients who are completely seizure-free since surgery) was reached in 64 patients (84%).
Mean follow-up duration in the DRE group was 134 months (range 24–281 months, SD 63 months). On the last available outcome (LAO) ILAE class 1 was achieved in 67 patients (88%) in the DRE group (see Table 3 for other ILAE classes). There were two patients (3%) with a new and permanent deficit who are not seizure-free in the DRE group (double losers).
Table 3. Follow-up/seizure outcome at last available follow-up
Average follow-up duration (months)
134 (SD 63)
141 (SD 64.4)
111 (SD 53.3)
Median follow-up duration (months)
At time of last available follow-up, 16 patients (80%) of patients with chronic epilepsy were classified as ILAE class 1. In the group of patients with sporadic seizures, 20 patients (91%) were categorized as ILAE class 1 at the time of last available follow-up. Three patients (15%) in the chronic epilepsy group were double losers.
Statistic comparison between lesionectomy (regardless whether with or without removal of the hemosiderotic rim) with more extended resections did not show any benefit for any of the applied surgical techniques regarding seizure outcome.
Regarding the localization of CCMs, statistical analysis reveals a better seizure outcome in the patients with temporal lobe CCMs (p = 0.02). It turned out that longer duration of symptoms in the groups with more long-standing types of epilepsy (DRE and chronic epilepsy group) was associated with worse seizure outcome (p = 0.01).
We present one of the largest single-center cohorts of patients with different types of CCM-associated seizures. The proportion of temporally located symptomatic CCM is 57% in the overall study population, which is more than in other studies (Zevgaridis et al., 1996; Moran et al., 1999), possibly due to the high proportion of DRE cases from the epilepsy surgery center. Nevertheless some studies focused only on temporal CCMs or predominantly included patients with temporally located CCMs (Hammen et al., 2007; Van Gompel et al., 2009; Kivelev et al., 2011). Therefore, the present patient cohort does differ somewhat from the previously published patient series. The limit of this study is its retrospective nature. Furthermore, we did not evaluate neuropsychological data to compare the potential subtle cognitive deficits after extended lesionectomy (Helmstaedter et al., 2002; Clusmann et al., 2004; Schramm & Clusmann, 2008). That is why extended lesionectomy as a standard cannot be proposed. The strengths of the study are the clear differentiation of the different types of epilepsy, the size of the study population, and the length of follow-up.
In other studies with a high percentage of CCMs in temporal localization, the predominant type of epilepsy was also intractable epilepsy (Stavrou et al., 2008; Van Gompel et al., 2009). In the two other types of epilepsy (chronic epilepsy and sporadic epilepsy), temporal localization was found in 32% and 29% of cases, respectively. The numbers correspond to the portion of temporal CCMs analyzed in a metaanalysis giving an overview about the regional distribution of symptomatic CCMs in the literature (Moran et al., 1999). Taken together, the results in this study show that temporally located CCMs predispose to develop DRE. Statistically, patients with extratemporal CCMs were not operated earlier than patients with temporal CCMs regarding duration of symptoms (149 vs. 112 months). Of interest, patients with temporally located CCMs showed superior seizure outcome compared with patients who have extratemporal CCM. This might be influenced by the fact that extended resections were performed more often in patients with temporally located CCMs.
Up to date, there are only a few studies available in which the authors did carefully subdivide the patient population according to seizure type, as described in a previous review (von der Brelie & Schramm, 2011). In those studies there is evidence that patients with DRE have a worse outcome compared with patients having sporadic seizures (Dodick et al., 1994; Zevgaridis et al., 1996; Rocamora et al., 2009). It has been shown previously that preoperative duration of epileptic symptoms is a prognostic factor for persisting postoperative seizures (Kim et al., 2011). Longer preoperative duration of symptoms was negatively correlated with postoperative seizure freedom in several studies (Cohen et al., 1995; Zevgaridis et al., 1996; Hammen et al., 2007).
For the two groups of more longstanding types of epilepsy (DRE and chronic epilepsy) it was shown that longer preoperative duration of symptoms was also correlated with slightly worse seizure outcome. The longer duration of seizure disorder found in the DRE group does not automatically lead to poorer seizure freedom rates. The lower rate of good ILAE grades for the chronic epilepsy group may result from the fact that detailed epileptologic workup was not done that frequently in patients with chronic epilepsy. The exact reasons for this are speculative, but apparently operative strategy in those cases was not tailored well enough to the epileptogenic zone. Possibly the invasive evaluation was not chosen as frequently because the severity of the epilepsy appeared not so grave. The good outcome in the DRE group is most likely due to the frequent use of invasive evaluation techniques. A likely explanation for the less frequent use of invasive evaluation in the other epilepsy groups is the impression that due to the lack of drug resistance a low level intensity in presurgical evaluation seemed to be adequate. One of the results of this study, therefore, is that in CCM-associated chronic epilepsy it may more frequently be justified to use invasive evaluation techniques to determine the extent of resection. Seizure-free rate of 88% in the DRE group is excellent, especially compared to results of epilepsy surgery in general, carried out in temporal, or extratemporal areas. Outcome in patients with sporadic seizures is best, however, reaching >90% in ILAE class 1. Because chances for ILAE class 1 outcome drop once patients have progressed from sporadic seizures to chronic epilepsy, it may be concluded that CCM-associated seizures should be treated by early surgery. Furthermore, it has been shown that nearly all CCM patients presenting with a first seizure will develop epilepsy within 5 years (Josephson et al., 2011). Because in this series the majority of cases presented with DRE it appears justified to argue that surgical intervention should be considered early, preferentially before seizures become drug resistant and in order to spare the unnecessary suffering of the patient from ongoing seizures and side effects of AED treatment.
Presurgical evaluation ideally delineates the extent of the epileptogenic zone. In this series the more intensive the evaluation and the more extensive the resection, the greater were the chances of postoperative seizure freedom. In temporal lobe epilepsy, invasive monitoring is usually indicated in cases of conflicting noninvasive data (Diehl & Luders, 2000). Previous data on the influence of invasive monitoring in epilepsy with single supratentorial CCMs is scarce; implantation techniques are rather used in epilepsy patients with multiple CCMs (Siegel et al., 2000; Rocamora et al., 2009). In the presented series, 23 patients received invasive monitoring. In all patients results were conclusive and a tailored epilepsy surgery strategy could be defined. ILAE class 1 was achieved in 18 (78%) of 23 patients with invasive monitoring. The procedure itself is safe; one patient developed a transient symptomatic subdural hematoma. In series dealing with temporal lobe epilepsy or other forms of epilepsy, outcome was worse in patients who underwent presurgical invasive monitoring. However, in this series, this was not the case; most likely invasive recording was not done to detect a completely unknown seizure focus, but just to outline the extent of resection of surrounding cortex in a well-known lesion.
It has been suspected that lesionectomy alone is not sufficient in surgical therapy of CCM-associated epilepsy. In the present series, most patients with DRE (87%) underwent operations with more extensive resections. This strategy leads to good seizure outcome. ILAE class 1 was achieved in 88%. In other series in which this subtype of epilepsy was defined and analyzed specifically, outcome rates vary. In the study of Baumann et al. (2007), resection type was lesionectomy including the hemosiderotic rim, and Engel class 1 outcome was 77% versus 65% in patients who underwent pure lesionectomy. In the series of Chang et al. (2009), patients with DRE underwent lesionectomy including the hemosiderotic rim, and outcome Engel class 1 was 77%. Chang et al. also used ECoG-guided extended resections and described a better outcome of 86% Engel class 1 for the ECoG group. In another series, van Gompel et al. also used ECoG guided extended lesionectomy. Outcome in the EcoG-treated group was Engel class 1 in 91% compared with 79% Engel class 1 outcome in patients who underwent only lesionectomy including the hemosiderotic rim. Therefore, the results of four studies underscore the value of detailed electrophysiologic workup followed by more extensive resections for DRE cases.
In cases with sporadic seizures the literature shows an excellent seizure outcome if lesionectomy is performed (Casazza et al., 1996; Ferroli et al., 2006). In the present series, seizure outcome in the subgroup of patients with sporadic seizures was 91% ILAE class 1. This result is in accordance with previous published data.
There is controversy about the potentially beneficial simultaneous resection of the hemosiderotic rim. Some studies favor this technique, showing a slightly superior outcome (Baumann et al., 2006). Other case series fail to show an outcome benefit correlated with removal of hemosiderin-stained tissue (Casazza et al., 1996; Zevgaridis et al., 1996). In this series the surgical technique was to remove the hemosiderotic fringe if feasible. It was removed in all but six cases, two of which were postoperatively categorized as ILAE class 4. Nevertheless, these numbers were too small to reach statistical significance.
The outcome in CCM-associated DRE can be very good if more extensive resections are used and if noninvasive and/or invasive presurgical epileptologic workup is used whenever indicated.
DRE was considerably more frequent in the temporal lobe, suggesting that temporal localization predisposes the development of DRE.
Seizure freedom rates were stable over a long period.
None of the authors has any conflict of interest to disclose. We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.