Neurosurg Focus 37 (3):E17, 2014 ©AANS, 2014
Seizure outcomes following radiosurgery for cerebral arteriovenous malformations Ching-Jen Chen, M.D.,1 Srinivas Chivukula, M.D., 2 Dale Ding, M.D.,1 Robert M. Starke, M.D., M.Sc.,1 Cheng-Chia Lee, M.D.,1,3 Chun-Po Yen, M.D.,1 Zhiyuan Xu, M.D.,1 and Jason P. Sheehan, M.D., Ph.D.1,4 Department of Neurological Surgery, University of Virginia Health System, Charlottesville, Virginia; Department of Neurological Surgery, University of Pittsburgh Medical Center, Pittsburgh, Pennsylvania; 3 Department of Neurosurgery, Neurological Institute, Taipei Veterans General Hospital, Taipei, Taiwan; and 4 Department of Radiation Oncology, University of Virginia Health System, Charlottesville, Virginia 1 2
Object. Seizures are a common presentation of cerebral arteriovenous malformations (AVMs). The authors evaluated the efficacy of stereotactic radiosurgery (SRS) for the management of seizures associated with AVMs and identified factors influencing seizure outcomes following SRS for AVMs. Methods. A systematic literature review was performed using PubMed. Studies selected for review were published in English, included at least 5 patients with both cerebral AVMs and presenting seizures treated with SRS, and provided post-SRS outcome data regarding obliteration of AVMs and/or seizures. Demographic, radiosurgical, radiological, and seizure outcome data were extracted and analyzed. All seizure outcomes were categorized as follows: 1) seizure free, 2) seizure improvement, 3) seizure unchanged, and 4) seizure worsened. Systematic statistical analysis was conducted to assess the effect of post-SRS AVM obliteration on seizure outcome. Results. Nineteen case series with a total of 3971 AVM patients were included for analysis. Of these, 28% of patients presented with seizures, and data for 997 patients with available seizure outcome data who met the inclusion criteria were evaluated. Of these, 437 (43.8%) patients achieved seizure-free status after SRS, and 530 (68.7%) of 771 patients with available data achieved seizure control (seizure freedom or seizure improvement) following SRS. Factors associated with improved seizure outcomes following SRS for AVMs were analyzed in 9 studies. Seizurefree status was achieved in 82% and 41.0% of patients with complete and incomplete AVM obliteration, respectively. Complete AVM obliteration offered superior seizure-free rates compared with incomplete AVM obliteration (OR 6.13; 95% CI 2.16–17.44; p = 0.0007). Conclusions. Stereotactic radiosurgery offers favorable seizure outcomes for AVM patients presenting with seizures. Improved seizure control is significantly more likely with complete AVM obliteration. (http://thejns.org/doi/abs/10.3171/2014.6.FOCUS1454)
Key Words • arteriovenous malformation • seizure • epilepsy • stereotactic radiosurgery • review
C
arteriovenous malformations (AVMs) are congenital vascular anomalies that have an estimated incidence of approximately 1 in 100,000 persons.3,9,30,35 Patients often present by the 3rd decade of life with hemorrhage, seizure, or neurological deficit.8,50 The most common presentation in patients harboring such vascular anomalies is hemorrhage, with the estimated risks ranging from 2% to 4% annually.11,27 Due to the significant morbidity and mortality associated with AVM rupture, the majority of AVM treatments and studies have focused on its natural incidence and the prevention of its erebral
Abbreviations used in this paper: AED = antiepileptic drug; AVM = arteriovenous malformation; GKRS = Gamma Knife radiosurgery; PBT = proton-beam therapy; SRS = stereotactic radiosurgery.
Neurosurg Focus / Volume 37 / September 2014
occurrence.25 In contrast, less attention has been given to seizures in patients with AVMs. Seizures occur in up to 57% of patients with cerebral AVMs.46 Among patients with unruptured AVMs, seizures are the most common clinical presentation.13 Despite antiepileptic drug (AED) therapy, seizure response remains variable for these patients.14,51,65 As disabling seizures can have detrimental effects on patients’ quality of life, reduction or elimination of seizures is crucial in the optimal management of these patients. Stereotactic radiosurgery (SRS) is a frequently employed modality for the treatment of AVMs; it offers a minimally invasive method of achieving AVM obliteration with rates that vary widely depending on multiple factors, including nidus volume, venous drainage pattern, prior embolization, and radiosurgical margin dose.13,42,47,59 1
C. J. Chen et al. To date, the goal of SRS in AVM treatment has been to eliminate the risk of hemorrhage by achieving complete obliteration of the nidus.13 However, the indications for SRS in the management of seizures associated with AVMs are not well defined, with relatively few studies investigating seizure outcomes following SRS.5,69 To evaluate the efficacy of SRS for the treatment of AVM-related seizures and identify factors associated with improved post-SRS seizure control, we performed a systematic review of the available literature regarding AVM seizure outcomes following SRS.
Methods Inclusion Criteria
Inclusion criteria for the studies were defined in an attempt to ensure a balance between a relatively homogeneous and the largest possible patient population. The following criteria were devised for inclusion in the final analysis: 1) the study must contain at least 5 patients with both an AVM and presenting seizures that were treated with SRS; 2) the study must include vascular malformations that were AVMs; 3) the study data must include post-SRS outcome data regarding AVM obliteration and/ or seizures; and 4) the language of the study must be in English. Case series consisting of fewer than 5 patients, studies pertaining to dural arteriovenous fistulas or cerebrovascular lesions other than AVMs (i.e., cavernous malformations), and studies in languages other than English were excluded from this analysis.
Literature Search
A systematic literature review was performed using PubMed. The literature search was performed using PubMed with the search term: “seizure OR epilepsy AND arteriovenous malformation AND radiosurgery.” The search included articles published from 1987 through 2013. The search yielded 198 articles, which were then screened by title and abstract. Twenty-two articles were selected from the initial screening process and were subject to further detailed reviewed for relevance and usable data matching our inclusion criteria. Of these, 3 studies were excluded for reasons including lack of or insufficient reported seizure outcomes and overlapping published patient data from the same institution in a more recent study. For the final quantitative analysis, we identified 19 series comprising 1104 patients with AVMs and seizures who underwent SRS.2,13,16,20,22,33,34,37,39,41,43,48,57,58,63,64,69,71,73 Figure 1 shows a flow chart of the review process.
Literature Review and Data Extraction
Demographic, radiosurgical, clinical, and radiological outcomes data were extracted from studies that met the inclusion criteria. Demographic data included the number of patients treated and the mean age and sex of all patients reported in each study, as well as those presenting with seizures. Other data obtained included Spetzler-Martin grade and history of AVM rupture.60 For patients with AVMs and seizures, follow-up duration and history of prior microsurgical resection and emboliza2
Fig. 1. Flow chart illustrating the systematic review process. The initial PubMed search yielded 198 articles; 22 articles were selected from the initial screening process, and these were further reviewed for relevance and usable data matching the inclusion criteria. Three studies were excluded for reasons including no or insufficient seizure outcomes reported and overlapping of published patient data from the same institution in a more recent study. In 19 case series with 3971 AVM patients treated with SRS, 1104 patients had concurrent seizures. Of these, 997 patients with radiological and/or clinical follow-up were included in this analysis.
tion were also recorded. For studies that distinguished patients with seizures following hemorrhage from those with hemorrhage-independent seizures, only data from patients with hemorrhage-independent seizures were included for analysis. Radiosurgical data reviewed included the type of SRS used (i.e., Gamma Knife radiosurgery [GKRS], proton-beam therapy [PBT], and LINAC-based radiosurgery) and treatment parameters when available. Data regarding seizure outcomes were also reviewed; all seizure outcomes were categorized into 4 groups as follows: 1) seizure free, 2) seizure improvement (decreased seizure frequency without seizure freedom), 3) seizure unchanged, and 4) seizure worsened. When available, the Engel Seizure Outcome Scale following SRS was noted.18 For studies in which only the Engel classification was reported, Class I was designated “seizure free,” Classes II and III were designated “seizure improvement,” and Class IV was reported as “seizure unchanged.” For the purpose of this review, the term “seizure control” encompassed Engel Classes I–III (patients who achieved seizure-free status and those with seizure improvement). For those with seizure-free status and/or seizure improvement, patients off AEDs following SRS were recorded Neurosurg Focus / Volume 37 / September 2014
Radiosurgery for AVMs: seizure outcomes when available. Rates of AVM obliteration were evaluated independently in each study by using a combination of angiography and MRI. Incomplete AVM obliteration included subtotal, partial, and no AVM obliteration. Factors associated with favorable seizure outcomes (seizurefree or seizure-control outcomes) following SRS were also noted. Statistical Analysis
All statistical analyses were performed using Review Manager (RevMan) version 5.2.8 (Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration, 2012). Seizure outcome data were extracted for patients with and without complete AVM obliteration from each study with available data. The Mantel-Haenszel test was used to compute the odds ratio. Under the assumption of possible clinical diversity and methodological variation between studies, the random-effects model was used and presented in this study. Study heterogeneity was detected using the chi-square and I2 test statistics. However, because the chi-square test lacks power when the number of studies is small, significant heterogeneity was considered to be present when both the chi-square value was within the 10% level of significance (p < 0.10) and the I2 value exceeded 50%. Possible reasons for variation across studies are addressed in the limitations section of this review. All statistical tests were 2-sided, and p < 0.05 was considered statistically significant.
Results AVM Radiosurgery Series Included for Analysis
Nineteen case series dating from 1991 to 2013 met the inclusion criteria; these studies contained 3971 AVM patients who were treated with SRS. Of these, 1104 (27.8%) patients presented with seizures; 997 of 1104 patients (90.3%) had radiological and/or clinical follow-up. The SRS modalities used in these studies included GKRS, LINAC, and PBT in 12, 4, and 2 series, respectively; 1 study used both GKRS and LINAC. Follow-up intervals ranged from 14 to 93 months. Table 1 summarizes the case series included in this review.
Seizure Outcome After SRS
Of the 997 patients with available follow-up, 437 (43.8%) patients achieved seizure-free status following SRS. In the 8 studies with reported AED use, 106 (53.5%) of 198 seizure-free patients were no longer being treated with AEDs at last follow-up. For 5 studies, only seizurefree outcomes were reported. Of the remaining 14 studies in which seizure improvement was reported, there were 771 patients with available follow-up, comprising 228 (29.6%) with seizure freedom and 242 (31.4%) with decreased seizure frequency, for a seizure control rate of 68.7%. Seizure status was unchanged in 210 (28.4%) of 739 patients in 13 studies with available data and worse in 20 (4.0%) of 494 patients in 8 studies. Table 1 shows seizure outcomes following SRS for AVMs in the included studies. When studies were analyzed using respective sei-
Neurosurg Focus / Volume 37 / September 2014
zure-free and seizure improvement percentages, the mean seizure-free rate was 53.4% (95% CI 39.6%–67.1%), and the mean seizure improvement rate was 26.6% (95% CI 15.2%–37.9%). A mean overall seizure control rate of 76.2% (95% CI 67.0%–85.3%) was achieved. The mean follow-up time for these studies was 45.6 months (range 14–93 months). However, for the few studies in which these data were reported, the time to achieving seizurefree outcome ranged from less than 3 months to 20.5 months. Obliteration and Other Factors Related to Seizure Outcome
Among the 9 studies with available radiological outcomes, complete AVM obliteration was achieved in 237 of 401 patients (59.1%, 95% CI 51.4%–66.8%). Seizure-free status was achieved in 176 of 215 patients with complete AVM obliteration (81.9%) in the 8 studies with available data, whereas seizure-free status was obtained in 59 of 144 patients with incomplete AVM obliteration (41.0%) in 7 studies with available data. Analysis of pooled data from the 7 studies with seizure-free outcomes for both complete and incomplete AVM obliteration demonstrated that complete obliteration offered superior seizure-free rates (OR 6.13; 95% CI 2.16–17.44; p = 0.0007). The analysis also demonstrated significant heterogeneity among the included studies (chi-square = 20.14; p = 0.003; I2 = 70%). Results of the analysis are summarized in Fig. 2. Factors associated with seizure outcomes following SRS for AVMs were analyzed in 9 studies and are summarized in Table 2. AVM obliteration was found to be significantly associated with improved seizure control in four studies.33,41,69,71 However, this association was found to be nonsignificant in 4 other studies.22,37,39,57 Relationships between seizure characteristics and seizure outcomes following SRS were also inconsistent; complex partial seizures with or without secondary generalization, generalized tonic-clonic seizures, infrequent and shortduration seizures, and seizure frequency have been associated with improved seizure control.16,33,39,57,71 In contrast, other studies have also found no significant relationship between seizure outcome and seizure type, duration, frequency, and pattern or age at seizure onset.16,37,39,57,71 The influence of hemorrhage on seizure outcome was also controversial. Two of the 3 studies that evaluated radiosurgery margin and/or maximum dose showed no significant association between these radiosurgical parameters and seizure outcome.39,57 Although in some studies an association was observed between improved seizure control and AVM location and size, many studies found no such association between seizure outcome and AVM characteristics (volume, size, drainage, location, and SpetzlerMartin grade). Patient age and sex were consistently not associated with seizure outcome.
Discussion
Approximately one-third of patients harboring AVMs present with seizures.16,21,43,73 In this systematic review, we observed an overall seizure incidence of 28% in studies that included AVM patients treated with SRS. Risk factors 3
4
Overall
227 247 160 72 100 315 462 424 49 285 26 38 246 69 318 161 444 164 164 3971
Authors & Year
Lunsford et al., 1991 Steiner et al., 1992 Sutcliffe et al., 1992 Gerszten et al., 1996 Eisenschenk et al., 1998 Kurita et al., 1998 Kida et al., 2000 Hoh et al., 2002 Nataf et al., 2003 Schäuble et al., 2004 Silander et al., 2004 Andrade-Souza et al., 2006 Lim et al., 2006 Zeiler et al., 2011 Hyun et al., 2012 Yang et al., 2012 Ding et al., 2013 Fokas et al., 2013 Wang et al., 2013 total
70 59 48 13 33 42 79 141 6 70 9 27 45 24 50 86 208 45 49 1104
w/ Seizures
No. of Patients
43 59 48 13 32 35 79 110 6 51 9 27 43 20 50 86 208 45 33 997
w/ Follow-Up
72 90 86 93 38
14 24 24 47 26 43 24 35 40 36 41 42 46
Follow-Up (mos) GKRS GKRS GKRS GKRS LINAC GKRS GKRS PBT LINAC GKRS PBT LINAC GKRS GKRS GKRS GKRS GKRS LINAC GKRS/LINAC
Treatment Modality
TABLE 1: Summary of SRS series and outcomes for patients with seizures and AVMs
22/33 237/401
60/86
15/27 16/33
9/14 21/35 22/57 68/110 4/6
No. of Patients w/ Complete AVM Obliteration/No. w/ AVMs 0 11 18 11 19 28 49 73 4 26 7 14 23 19 33 66 16 0 20 437 (43.8%)
242 (32.7%)
210 (28.4%)
93 14
2 1 8 1
0 11 10 0
89 28
2 13 22
20 18 16 0
Unchanged
22 30 11 2 6 4 17 12
Seizure Free Improvement
20 (4.0%)
10 3
2
1
1 0 3 0
Worsened
Seizure Status After SRS (no. of patients)
C. J. Chen et al.
Neurosurg Focus / Volume 37 / September 2014
Radiosurgery for AVMs: seizure outcomes
Fig. 2. Forest plot of the odds ratio of seizure-free outcomes for patients with complete and incomplete AVM obliteration. The estimated odds ratio and 95% CI of each included study is represented by the center of the squares and the horizontal line, respectively. The summary odds ratio and 95% CI are shown in bold and are represented by the black diamond. Tests of heterogeneity and overall effect are given below the summary statistics. M-H = Mantel-Haenszel; Oblit. = obliteration.
for seizure presentation in these patients have been previously documented and include younger age, male sex, lobar or cortical AVM location, and large AVM size.12,21,23, 33,40,44,46,52,61,67,68 Other AVM characteristics associated with seizures, such as frontal or temporal topography, superficial venous drainage, and arterial border-zone AVM location, have also been reported.21,23,33,40,52,62,68 While defining factors for seizure development in patients with AVMs may be important for risk stratification, analyzing seizure outcome following AVM treatment will help direct future therapy. Although seizures are the second most common presentation in patients with AVMs, seizure control has not been the primary objective of AVM treatment.31 Current treatment options for patients with AVMs in-
clude microsurgery, embolization, and SRS. Of these, SRS has become an increasingly popular therapeutic modality. This systematic review evaluated the efficacy of SRS in treating patients with AVM-associated seizures in 19 studies. The seizure-free rates following SRS varied widely, ranging from 0% to 95%, with a mean of seizure-free rate of 53.4% in the 19 identified studies. In addition, approximately half of the seizure-free patients were weaned off of AEDs after SRS. After a 2-year seizure-free status, there is general consensus for a slow withdrawal of AEDs, but the interval to freedom from AEDs was reported in only 1 study.39,41,71 Yang et al. observed a median time of 4 years (95% CI 2.8–5.2 years) to achieving AED cessation following SRS.71 SRS seems to offer reasonable seizure-control
TABLE 2: Factors associated with seizure outcomes following SRS observed in different studies Authors & Year
Factors Associated w/ Improved Seizure Outcomes
Gerszten et al., no hemorrhage, higher-grade AVM 1996 Eisenschenk et al., complex partial seizure w/ or w/o secondary generalization, 1998 generalized tonic-clonic seizure w/o preceding partial sei zures, AVM located in centrum or frontal regions Kurita et al., 1998 infrequent seizures prior to SRS, short duration (≤6 mos) of seizure history Kida et al., 2000 Hoh et al., 2002 Schäuble et al., 2004 Lim et al., 2006 Yang et al., 2012
Wang et al., 2013
Factors Not Associated w/ Improved Seizure Outcomes AVM obliteration
simple partial seizure (unfavorable), AVM in temporal lobe exclusively or contiguous to temporal lobe (unfavorable), AVM size age & sex; seizure type & age at seizure onset; AVM side, loca tion, drainage, nidus diameter, nidus vol; radiosurgery central & margin dose; AVM obliteration seizure as initial symptom (compared to following hemorrhage), AVM size, AVM obliteration, seizure pattern radiosurgery margin dose
Seizure outcomes following radiosurgery for cerebral arteriovenous malformations Ching-Jen Chen, M.D.,1 Srinivas Chivukula, M.D., 2 Dale Ding, M.D.,1 Robert M. Starke, M.D., M.Sc.,1 Cheng-Chia Lee, M.D.,1,3 Chun-Po Yen, M.D.,1 Zhiyuan Xu, M.D.,1 and Jason P. Sheehan, M.D., Ph.D.1,4 Department of Neurological Surgery, University of Virginia Health System, Charlottesville, Virginia; Department of Neurological Surgery, University of Pittsburgh Medical Center, Pittsburgh, Pennsylvania; 3 Department of Neurosurgery, Neurological Institute, Taipei Veterans General Hospital, Taipei, Taiwan; and 4 Department of Radiation Oncology, University of Virginia Health System, Charlottesville, Virginia 1 2
Object. Seizures are a common presentation of cerebral arteriovenous malformations (AVMs). The authors evaluated the efficacy of stereotactic radiosurgery (SRS) for the management of seizures associated with AVMs and identified factors influencing seizure outcomes following SRS for AVMs. Methods. A systematic literature review was performed using PubMed. Studies selected for review were published in English, included at least 5 patients with both cerebral AVMs and presenting seizures treated with SRS, and provided post-SRS outcome data regarding obliteration of AVMs and/or seizures. Demographic, radiosurgical, radiological, and seizure outcome data were extracted and analyzed. All seizure outcomes were categorized as follows: 1) seizure free, 2) seizure improvement, 3) seizure unchanged, and 4) seizure worsened. Systematic statistical analysis was conducted to assess the effect of post-SRS AVM obliteration on seizure outcome. Results. Nineteen case series with a total of 3971 AVM patients were included for analysis. Of these, 28% of patients presented with seizures, and data for 997 patients with available seizure outcome data who met the inclusion criteria were evaluated. Of these, 437 (43.8%) patients achieved seizure-free status after SRS, and 530 (68.7%) of 771 patients with available data achieved seizure control (seizure freedom or seizure improvement) following SRS. Factors associated with improved seizure outcomes following SRS for AVMs were analyzed in 9 studies. Seizurefree status was achieved in 82% and 41.0% of patients with complete and incomplete AVM obliteration, respectively. Complete AVM obliteration offered superior seizure-free rates compared with incomplete AVM obliteration (OR 6.13; 95% CI 2.16–17.44; p = 0.0007). Conclusions. Stereotactic radiosurgery offers favorable seizure outcomes for AVM patients presenting with seizures. Improved seizure control is significantly more likely with complete AVM obliteration. (http://thejns.org/doi/abs/10.3171/2014.6.FOCUS1454)
Key Words • arteriovenous malformation • seizure • epilepsy • stereotactic radiosurgery • review
C
arteriovenous malformations (AVMs) are congenital vascular anomalies that have an estimated incidence of approximately 1 in 100,000 persons.3,9,30,35 Patients often present by the 3rd decade of life with hemorrhage, seizure, or neurological deficit.8,50 The most common presentation in patients harboring such vascular anomalies is hemorrhage, with the estimated risks ranging from 2% to 4% annually.11,27 Due to the significant morbidity and mortality associated with AVM rupture, the majority of AVM treatments and studies have focused on its natural incidence and the prevention of its erebral
Abbreviations used in this paper: AED = antiepileptic drug; AVM = arteriovenous malformation; GKRS = Gamma Knife radiosurgery; PBT = proton-beam therapy; SRS = stereotactic radiosurgery.
Neurosurg Focus / Volume 37 / September 2014
occurrence.25 In contrast, less attention has been given to seizures in patients with AVMs. Seizures occur in up to 57% of patients with cerebral AVMs.46 Among patients with unruptured AVMs, seizures are the most common clinical presentation.13 Despite antiepileptic drug (AED) therapy, seizure response remains variable for these patients.14,51,65 As disabling seizures can have detrimental effects on patients’ quality of life, reduction or elimination of seizures is crucial in the optimal management of these patients. Stereotactic radiosurgery (SRS) is a frequently employed modality for the treatment of AVMs; it offers a minimally invasive method of achieving AVM obliteration with rates that vary widely depending on multiple factors, including nidus volume, venous drainage pattern, prior embolization, and radiosurgical margin dose.13,42,47,59 1
C. J. Chen et al. To date, the goal of SRS in AVM treatment has been to eliminate the risk of hemorrhage by achieving complete obliteration of the nidus.13 However, the indications for SRS in the management of seizures associated with AVMs are not well defined, with relatively few studies investigating seizure outcomes following SRS.5,69 To evaluate the efficacy of SRS for the treatment of AVM-related seizures and identify factors associated with improved post-SRS seizure control, we performed a systematic review of the available literature regarding AVM seizure outcomes following SRS.
Methods Inclusion Criteria
Inclusion criteria for the studies were defined in an attempt to ensure a balance between a relatively homogeneous and the largest possible patient population. The following criteria were devised for inclusion in the final analysis: 1) the study must contain at least 5 patients with both an AVM and presenting seizures that were treated with SRS; 2) the study must include vascular malformations that were AVMs; 3) the study data must include post-SRS outcome data regarding AVM obliteration and/ or seizures; and 4) the language of the study must be in English. Case series consisting of fewer than 5 patients, studies pertaining to dural arteriovenous fistulas or cerebrovascular lesions other than AVMs (i.e., cavernous malformations), and studies in languages other than English were excluded from this analysis.
Literature Search
A systematic literature review was performed using PubMed. The literature search was performed using PubMed with the search term: “seizure OR epilepsy AND arteriovenous malformation AND radiosurgery.” The search included articles published from 1987 through 2013. The search yielded 198 articles, which were then screened by title and abstract. Twenty-two articles were selected from the initial screening process and were subject to further detailed reviewed for relevance and usable data matching our inclusion criteria. Of these, 3 studies were excluded for reasons including lack of or insufficient reported seizure outcomes and overlapping published patient data from the same institution in a more recent study. For the final quantitative analysis, we identified 19 series comprising 1104 patients with AVMs and seizures who underwent SRS.2,13,16,20,22,33,34,37,39,41,43,48,57,58,63,64,69,71,73 Figure 1 shows a flow chart of the review process.
Literature Review and Data Extraction
Demographic, radiosurgical, clinical, and radiological outcomes data were extracted from studies that met the inclusion criteria. Demographic data included the number of patients treated and the mean age and sex of all patients reported in each study, as well as those presenting with seizures. Other data obtained included Spetzler-Martin grade and history of AVM rupture.60 For patients with AVMs and seizures, follow-up duration and history of prior microsurgical resection and emboliza2
Fig. 1. Flow chart illustrating the systematic review process. The initial PubMed search yielded 198 articles; 22 articles were selected from the initial screening process, and these were further reviewed for relevance and usable data matching the inclusion criteria. Three studies were excluded for reasons including no or insufficient seizure outcomes reported and overlapping of published patient data from the same institution in a more recent study. In 19 case series with 3971 AVM patients treated with SRS, 1104 patients had concurrent seizures. Of these, 997 patients with radiological and/or clinical follow-up were included in this analysis.
tion were also recorded. For studies that distinguished patients with seizures following hemorrhage from those with hemorrhage-independent seizures, only data from patients with hemorrhage-independent seizures were included for analysis. Radiosurgical data reviewed included the type of SRS used (i.e., Gamma Knife radiosurgery [GKRS], proton-beam therapy [PBT], and LINAC-based radiosurgery) and treatment parameters when available. Data regarding seizure outcomes were also reviewed; all seizure outcomes were categorized into 4 groups as follows: 1) seizure free, 2) seizure improvement (decreased seizure frequency without seizure freedom), 3) seizure unchanged, and 4) seizure worsened. When available, the Engel Seizure Outcome Scale following SRS was noted.18 For studies in which only the Engel classification was reported, Class I was designated “seizure free,” Classes II and III were designated “seizure improvement,” and Class IV was reported as “seizure unchanged.” For the purpose of this review, the term “seizure control” encompassed Engel Classes I–III (patients who achieved seizure-free status and those with seizure improvement). For those with seizure-free status and/or seizure improvement, patients off AEDs following SRS were recorded Neurosurg Focus / Volume 37 / September 2014
Radiosurgery for AVMs: seizure outcomes when available. Rates of AVM obliteration were evaluated independently in each study by using a combination of angiography and MRI. Incomplete AVM obliteration included subtotal, partial, and no AVM obliteration. Factors associated with favorable seizure outcomes (seizurefree or seizure-control outcomes) following SRS were also noted. Statistical Analysis
All statistical analyses were performed using Review Manager (RevMan) version 5.2.8 (Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration, 2012). Seizure outcome data were extracted for patients with and without complete AVM obliteration from each study with available data. The Mantel-Haenszel test was used to compute the odds ratio. Under the assumption of possible clinical diversity and methodological variation between studies, the random-effects model was used and presented in this study. Study heterogeneity was detected using the chi-square and I2 test statistics. However, because the chi-square test lacks power when the number of studies is small, significant heterogeneity was considered to be present when both the chi-square value was within the 10% level of significance (p < 0.10) and the I2 value exceeded 50%. Possible reasons for variation across studies are addressed in the limitations section of this review. All statistical tests were 2-sided, and p < 0.05 was considered statistically significant.
Results AVM Radiosurgery Series Included for Analysis
Nineteen case series dating from 1991 to 2013 met the inclusion criteria; these studies contained 3971 AVM patients who were treated with SRS. Of these, 1104 (27.8%) patients presented with seizures; 997 of 1104 patients (90.3%) had radiological and/or clinical follow-up. The SRS modalities used in these studies included GKRS, LINAC, and PBT in 12, 4, and 2 series, respectively; 1 study used both GKRS and LINAC. Follow-up intervals ranged from 14 to 93 months. Table 1 summarizes the case series included in this review.
Seizure Outcome After SRS
Of the 997 patients with available follow-up, 437 (43.8%) patients achieved seizure-free status following SRS. In the 8 studies with reported AED use, 106 (53.5%) of 198 seizure-free patients were no longer being treated with AEDs at last follow-up. For 5 studies, only seizurefree outcomes were reported. Of the remaining 14 studies in which seizure improvement was reported, there were 771 patients with available follow-up, comprising 228 (29.6%) with seizure freedom and 242 (31.4%) with decreased seizure frequency, for a seizure control rate of 68.7%. Seizure status was unchanged in 210 (28.4%) of 739 patients in 13 studies with available data and worse in 20 (4.0%) of 494 patients in 8 studies. Table 1 shows seizure outcomes following SRS for AVMs in the included studies. When studies were analyzed using respective sei-
Neurosurg Focus / Volume 37 / September 2014
zure-free and seizure improvement percentages, the mean seizure-free rate was 53.4% (95% CI 39.6%–67.1%), and the mean seizure improvement rate was 26.6% (95% CI 15.2%–37.9%). A mean overall seizure control rate of 76.2% (95% CI 67.0%–85.3%) was achieved. The mean follow-up time for these studies was 45.6 months (range 14–93 months). However, for the few studies in which these data were reported, the time to achieving seizurefree outcome ranged from less than 3 months to 20.5 months. Obliteration and Other Factors Related to Seizure Outcome
Among the 9 studies with available radiological outcomes, complete AVM obliteration was achieved in 237 of 401 patients (59.1%, 95% CI 51.4%–66.8%). Seizure-free status was achieved in 176 of 215 patients with complete AVM obliteration (81.9%) in the 8 studies with available data, whereas seizure-free status was obtained in 59 of 144 patients with incomplete AVM obliteration (41.0%) in 7 studies with available data. Analysis of pooled data from the 7 studies with seizure-free outcomes for both complete and incomplete AVM obliteration demonstrated that complete obliteration offered superior seizure-free rates (OR 6.13; 95% CI 2.16–17.44; p = 0.0007). The analysis also demonstrated significant heterogeneity among the included studies (chi-square = 20.14; p = 0.003; I2 = 70%). Results of the analysis are summarized in Fig. 2. Factors associated with seizure outcomes following SRS for AVMs were analyzed in 9 studies and are summarized in Table 2. AVM obliteration was found to be significantly associated with improved seizure control in four studies.33,41,69,71 However, this association was found to be nonsignificant in 4 other studies.22,37,39,57 Relationships between seizure characteristics and seizure outcomes following SRS were also inconsistent; complex partial seizures with or without secondary generalization, generalized tonic-clonic seizures, infrequent and shortduration seizures, and seizure frequency have been associated with improved seizure control.16,33,39,57,71 In contrast, other studies have also found no significant relationship between seizure outcome and seizure type, duration, frequency, and pattern or age at seizure onset.16,37,39,57,71 The influence of hemorrhage on seizure outcome was also controversial. Two of the 3 studies that evaluated radiosurgery margin and/or maximum dose showed no significant association between these radiosurgical parameters and seizure outcome.39,57 Although in some studies an association was observed between improved seizure control and AVM location and size, many studies found no such association between seizure outcome and AVM characteristics (volume, size, drainage, location, and SpetzlerMartin grade). Patient age and sex were consistently not associated with seizure outcome.
Discussion
Approximately one-third of patients harboring AVMs present with seizures.16,21,43,73 In this systematic review, we observed an overall seizure incidence of 28% in studies that included AVM patients treated with SRS. Risk factors 3
4
Overall
227 247 160 72 100 315 462 424 49 285 26 38 246 69 318 161 444 164 164 3971
Authors & Year
Lunsford et al., 1991 Steiner et al., 1992 Sutcliffe et al., 1992 Gerszten et al., 1996 Eisenschenk et al., 1998 Kurita et al., 1998 Kida et al., 2000 Hoh et al., 2002 Nataf et al., 2003 Schäuble et al., 2004 Silander et al., 2004 Andrade-Souza et al., 2006 Lim et al., 2006 Zeiler et al., 2011 Hyun et al., 2012 Yang et al., 2012 Ding et al., 2013 Fokas et al., 2013 Wang et al., 2013 total
70 59 48 13 33 42 79 141 6 70 9 27 45 24 50 86 208 45 49 1104
w/ Seizures
No. of Patients
43 59 48 13 32 35 79 110 6 51 9 27 43 20 50 86 208 45 33 997
w/ Follow-Up
72 90 86 93 38
14 24 24 47 26 43 24 35 40 36 41 42 46
Follow-Up (mos) GKRS GKRS GKRS GKRS LINAC GKRS GKRS PBT LINAC GKRS PBT LINAC GKRS GKRS GKRS GKRS GKRS LINAC GKRS/LINAC
Treatment Modality
TABLE 1: Summary of SRS series and outcomes for patients with seizures and AVMs
22/33 237/401
60/86
15/27 16/33
9/14 21/35 22/57 68/110 4/6
No. of Patients w/ Complete AVM Obliteration/No. w/ AVMs 0 11 18 11 19 28 49 73 4 26 7 14 23 19 33 66 16 0 20 437 (43.8%)
242 (32.7%)
210 (28.4%)
93 14
2 1 8 1
0 11 10 0
89 28
2 13 22
20 18 16 0
Unchanged
22 30 11 2 6 4 17 12
Seizure Free Improvement
20 (4.0%)
10 3
2
1
1 0 3 0
Worsened
Seizure Status After SRS (no. of patients)
C. J. Chen et al.
Neurosurg Focus / Volume 37 / September 2014
Radiosurgery for AVMs: seizure outcomes
Fig. 2. Forest plot of the odds ratio of seizure-free outcomes for patients with complete and incomplete AVM obliteration. The estimated odds ratio and 95% CI of each included study is represented by the center of the squares and the horizontal line, respectively. The summary odds ratio and 95% CI are shown in bold and are represented by the black diamond. Tests of heterogeneity and overall effect are given below the summary statistics. M-H = Mantel-Haenszel; Oblit. = obliteration.
for seizure presentation in these patients have been previously documented and include younger age, male sex, lobar or cortical AVM location, and large AVM size.12,21,23, 33,40,44,46,52,61,67,68 Other AVM characteristics associated with seizures, such as frontal or temporal topography, superficial venous drainage, and arterial border-zone AVM location, have also been reported.21,23,33,40,52,62,68 While defining factors for seizure development in patients with AVMs may be important for risk stratification, analyzing seizure outcome following AVM treatment will help direct future therapy. Although seizures are the second most common presentation in patients with AVMs, seizure control has not been the primary objective of AVM treatment.31 Current treatment options for patients with AVMs in-
clude microsurgery, embolization, and SRS. Of these, SRS has become an increasingly popular therapeutic modality. This systematic review evaluated the efficacy of SRS in treating patients with AVM-associated seizures in 19 studies. The seizure-free rates following SRS varied widely, ranging from 0% to 95%, with a mean of seizure-free rate of 53.4% in the 19 identified studies. In addition, approximately half of the seizure-free patients were weaned off of AEDs after SRS. After a 2-year seizure-free status, there is general consensus for a slow withdrawal of AEDs, but the interval to freedom from AEDs was reported in only 1 study.39,41,71 Yang et al. observed a median time of 4 years (95% CI 2.8–5.2 years) to achieving AED cessation following SRS.71 SRS seems to offer reasonable seizure-control
TABLE 2: Factors associated with seizure outcomes following SRS observed in different studies Authors & Year
Factors Associated w/ Improved Seizure Outcomes
Gerszten et al., no hemorrhage, higher-grade AVM 1996 Eisenschenk et al., complex partial seizure w/ or w/o secondary generalization, 1998 generalized tonic-clonic seizure w/o preceding partial sei zures, AVM located in centrum or frontal regions Kurita et al., 1998 infrequent seizures prior to SRS, short duration (≤6 mos) of seizure history Kida et al., 2000 Hoh et al., 2002 Schäuble et al., 2004 Lim et al., 2006 Yang et al., 2012
Wang et al., 2013
Factors Not Associated w/ Improved Seizure Outcomes AVM obliteration
simple partial seizure (unfavorable), AVM in temporal lobe exclusively or contiguous to temporal lobe (unfavorable), AVM size age & sex; seizure type & age at seizure onset; AVM side, loca tion, drainage, nidus diameter, nidus vol; radiosurgery central & margin dose; AVM obliteration seizure as initial symptom (compared to following hemorrhage), AVM size, AVM obliteration, seizure pattern radiosurgery margin dose
