Seizure predictors and outcome after Onyx embolization in patients with brain arteriovenous malformations

Abstract

Background

Brain arteriovenous malformation (bAVM)-related epilepsy can significantly affect patient quality of life. We aimed to identify the factors associated with seizures occurrence and evaluate the long-term outcome following Onyx embolization in bAVM patients.

Methods

Between July 2014 and July 2016, 239 consecutive patients underwent treatment for bAVMs in our institute and were respectively analyzed. Demographics, seizure status and bAVM morphologic characteristics were recorded. Modified Engel classification was used to evaluate the long-term seizure outcomes.

Results

Of 239 bAVM patients, 68 (28.5%) initially presented with seizures. Seizure occurrence was associated with cerebral hemorrhage history, frontal-temporal location and arterial borderzone location. Of the 37 patients who presented with initial seizures and were treated with Onyx embolization, 23 (62.2%) were treated with antiepileptic drugs (AEDs) before Onyx embolization. At the last follow-up visit, 19 (51.4%) of the 37 patients reached modified Engel class I outcome. Of the 23 patients who had ever been treated with AEDs, 12 (52.2%) were still taking AEDs at the last follow-up visit. Single-factor analysis showed that arterial borderzone location was significantly correlated with higher modified Engel class outcome (p = 0.046).

Conclusion

Patients with bAVM hemorrhage history, frontal-temporal location and arterial borderzone location were associated with seizure occurrence. Seizure-free status was not obtained in AVM patients with arterial borderzone after embolization, though it may have benefits in other ways. The seizure-free mechanism of bAVM with Onyx embolization is worth further study.

Introduction

Epilepsy is the second most common clinical symptom associated with brain arteriovenous malformations (bAVMs).¹ It has been estimated that between 29% and 43% of patients with bAVMs will experience seizures as part of their symptomatology.^2–7 Uncontrolled bAVM-related epilepsy can significantly affect patient quality of life. The main aim of treating bAVMs has been to decrease the hemorrhage of the nidus, but the epileptic seizure also needs to be controlled. Previous reports indicated that surgery and stereotactic radiosurgery (SRS) were of benefit for seizure control. However, patients who did not experience seizures before treatment may exhibit new-onset seizures after surgery and SRS.^2,5,6,8 Previous studies mainly involved in surgery and SRS, and only scant data exist on seizure outcome of bAVM patients following embolization.^9,10 That few studies focused on seizure outcome after AVM embolization may be because embolization was considered only as an adjunct to surgery or SRS.

In the current study, we aimed to analyze the demographic and morphologic characteristics of patients with bAVM-related seizures at initial presentation and explore the long-term seizure outcome of bAVM patients treated with Onyx embolization.

Methods

Study population and data collection

Data from patients diagnosed with bAVMs from July 2014 to July 2016 at our institute were retrospectively analyzed. This study was approved by the ethics committee of our hospital, and all participants provided their informed consent. For analyzing the risk factors of AVM-related seizure, we included that patients were diagnosed with bAVMs by digital subtraction angiography (DSA), and excluded that patients with unobtainable medical record data and cerebrovascular disease comorbidities (e.g. aneurysms, arteriovenous fistulas, moyamoya disease). To further explorie the risk factors of seizure-free status of AVM patients after Onyx embolization, we included patients who treated with Onyx embolization as a main part of their overall management and who patients had a follow-up examination at least 12 months after embolization, and excluded patients who were treated with surgical resection (because of the possible association between surgical resection and postoperative seizures), eight patients who were treated with SRS alone, and patients who were treated with NBCA embolization. A flowchart depicting the study design and patient selection is shown in Figure 1.

Figure 1.

Flowchart depicting study design and patient selection.

bAVM: brain arteriovenous malformation; SRS: stereotactic radiosurgery.

Definition of variables

Demographic data were gathered from electronic medical records and included age, sex, age at diagnosis, and initial clinical manifestation. Morphologic variables including size, location, side, feeding arteries and venous drainage patterns were obtained by DSA and/or magnetic resonance imaging (MRI) and analyzed by two interventional neuroradiologists.

Venous drainage patterns were categorized as angiographic drainage into the superficial veins or sinuses, exclusive of drainage into the deep venous system, or combined superficial and deep venous drainage. An arterial borderzone location was labeled as positive when the bAVMs were supplied by branches of at least two of the individual major circle of Willis arteries (i.e. the anterior and middle, middle and posterior, anterior and posterior, or anterior, middle, and posterior cerebral arteries).¹¹ Spetzler-Martin (S-M) grades were used to evaluate bAVM features.¹²

Epilepsy was classified according to the International League Against Epilepsy criteria.¹³ A modified version of the Engel classification system was used to stratify postoperative outcomes into patients who had zero or only one postoperative seizure (modified Engel class I outcome) and those who experienced multiple postoperative seizures (modified Engel class II–IV outcome) as of their last follow-up.¹⁴ “Seizure-free” patients were defined as individuals who presented with seizures initially and showed a modified Engel class I outcome at least 12 months after treatment. “Postoperation time” was defined as a period of ≥12 consecutive months from treatment to the last follow-up. Persistent seizures despite an adequate trial of two or more appropriate antiepileptic drugs (AEDs) were classified as drug-resistant epilepsy (DRE). Patients were classified as having sporadic seizures (SPS) if they had had one or two seizures.

Treatment criteria

In our institute, bAVM patients were treated with surgery, embolization, SRS and combinations thereof. Therapeutic options were discussed by a multidisciplinary team of neurosurgeons, neurologists, interventional neuroradiologists and radiation oncologists. Patients who underwent embolization as a main part of their overall management were included in this study. Endovascular therapy is the method of choice for small AVMs when complete endovascular obliteration seems feasible. Targeted embolization was used to eliminate AVM high-risk features (e.g. perinidus aneurysms, fistula) that predispose to hemorrhage. Palliative embolization was used for AVM patients with seizure, progressive neurological deficits, mental decline and persistent headache.

Embolization technique

bAVM embolization was performed under general anesthesia. The femoral artery was punctured and a 6F femoral artery sheath was inserted. According to the location of the bAVM, a 6-Fr guiding catheter was inserted into the appropriate artery. A flow-directed microcatheter (Marathon, ev3) was guided by microguidewire into the feeding artery under roadmap guidance. In case of selective angiography, the vascular architecture of the nidus and its feeding artery, as well as a suitable location for the microcatheter, were a priori verified. Saline was used to flush the microcatheter, and then the microcatheter lumen was filled with dimethyl sulfoxide. We generally used Onyx-18 (Covidien) as the embolic agent. The “reflux-hold-reinjection” technology was used, and glue reflux or diffusion into the drainage vein was avoided. After the procedure was completed, the microcatheter was gently removed.

Follow-up

Seizure outcomes were assessed at follow-up visits after the last treatment, and patient AED status was also obtained. In the final follow-up, seizure outcome and postoperative AED status were obtained by telephone. In this study, modified Engel class I was defined as favorable seizure outcomes, and a modified Engel class II–IV outcome was defined as poor outcome. DSA was performed after six to 12 months in patients with embolization. MRI was performed every six months for patients after endovascular embolization and SRS therapy, and DSA was usually performed two years after SRS.

Statistical analysis

Continuous variables are presented as mean ± standard deviation, and categorical variables as absolute and relative frequency (%). To analyze the factors associated with seizure occurrence before embolization and seizure control after embolization, Pearson χ² or the Fisher exact test was used for categorical variables, and the Student’s t test or Mann-Whitney U test was used for continuous variables. Variables with p < 0.15 on the univariate analysis were included in the multivariate analysis. Results are presented as odds ratios (OR) with a 95% confidence interval (CI). All p values are based on two-sided tests, and p < 0.05 was considered significant. All statistical analyses were carried out in Excel 2010 (Microsoft, Seattle, WA, USA) and SPSS 22 (Armonk, NY, USA).

Results

Factors associated with pretreatment seizures

A total of 239 patients diagnosed with bAVM were enrolled in our cohort, including 130 (54.4%) males and 109 (45.6%) females. Of the 239 patients analyzed, 68 (28.5%) initially presented with a history of seizures. The mean age at seizure onset was 24.8 ± 12.7 years (range, 3–64 years). Of the 68 patients who presented with a history of seizures, 28 (41.2%) presented with a history of simple partial seizures, four (5.9%) presented with complex partial seizures and 36 (52.9%) presented with generalized seizures. Seizure history was found to be significantly associated with the occurrence of bAVM hemorrhage history (p < 0.001), frontal-temporal location (p = 0.017), size ≥ 6 cm (p = 0.001), superficial venous drainage (p = 0.015) and arterial borderzone location (p = 0.007). However, sex, age, S-M grade, eloquence, nidus diffuseness and nidus side were not significantly related to seizure history preceding embolization. In multivariate analyses, seizure occurrence was significantly associated with a history of bAVM hemorrhage (OR: 10.548; 95% CI: 5.034–22.102; p < 0.001), frontal-temporal location (OR: 2.476; 95% CI: 1.181–5.190; p = .016) and arterial borderzone location (OR: 2.241; 95% CI: 1.082–4.641; p = .030) (Table 1).

Table 1.

Factors associated with preoperative seizure occurrence of all arteriovenous malformation patients.

Characteristics	Patients with seizure	Patients without seizure	Univariate analysis	Multivariate analysis
			p	OR (95% CI for OR)	p
Age at initial seizure	24.8 ± 12.7
Patients no.	68 (28.5)	171 (71.5)
Sex (male)	39 (57.4)	91 (53.2)	0.562
Age (mean ± SD)	29.8 ± 13.4	27.7 ± 12.8	0.257
Cerebral hemorrhage history	12 (17.6)	123 (71.9)	< 0.001	10.548 (5.034–22.102)	< 0.001
AVM location
Frontal-temporal	30 (55.9)	48 (28.1)	0.017	2.476 (1.181–5.190)	0.016
other	38 (44.1)	123 (71.9)		Reference
Size, cm
D < 6 cm	52 (76.5)	157 (91.8)	0.001	Reference	0.093
6 cm ≤ D	16 (23.5)	14 (8.2)		2.219 (0.877–5.619)
Eloquence	49 (94.1)	129 (86.5)	0.589
Nidus diffuseness	40 (58.8)	105 (61.4)	0.713
Venous drainage patterns
Superficial drainage	49 (72.1)	94 (55.0)	0.015	1.731 (0.826–3.628)	0.146
Deep drainage	19 (27.9)	77 (45.0)		Reference
S-M grade, n (%)
I	7 (10.3)	18 (10.5)	0.386
II	17 (25.0)	52 (30.4)
III	26 (38.2)	70 (40.9)
IV	13 (19.1)	27 (15.8)
V	5 (7.4)	4 (2.3)
Side
Left	32 (47.1)	71 (41.5)	0.140	0.906 (0.482–1.703)	0.760
Right	36 (52.9)	91 (53.2)
Crosses midline	0 (0)	9 (5.3)
Arterial border zone location
Yes	48 (70.6)	88 (51.5)	0.007	2.241 (1.082–4.641)	0.030
No	20 (29.4)	83 (48.5)		Reference
Predominant seizure type
Simple partial seizures	28 (41.2)
Complex partial seizures	4 (5.9)
generalized seizures	36 (52.9)

AVM: arteriovenous malformation; CI: confidence interval; D: diameter; OR: odds ratio; SD: standard deviation; S-M: Spetzler-Martin.

Outcomes in patients with pretreatment seizure after Onyx embolization

A total of 37 patients with initial seizures were treated with Onyx embolization as the main part of their overall management after excluding the patients treated with surgical resection and SRS in our cohort. There were 19 (51.4%) patients treated with embolization alone, and 18 (48.6%) treated with embolization and SRS therapy. Intracranial hemorrhage occurred in five patients before embolization. Total obliteration of the AVM was achieved in 19 patients (51.4%). One AVM ruptured during embolization, one patient suffered AVM rupture one day after embolization, and two patients developed transient neurological deficits after embolization. None of the 37 patients died postoperatively.

Of the 37 patients who presented with a history of seizures, 19 patients attained seizure-free status. Seizure-free rate was worse in DRE cases (37.5%) compared with SPS (57.9%) and other seizure type (50.0%). Twenty-five patients with generalized seizures acquired a 48.0% seizure-free rate. Twelve patients with simple partial seizure acquired a 58.3% seizure-free rate. There were 18 patients with bAVMs located on the frontal-temporal lobe, and eight patients (44.4%) of them with seizure free. The rates of seizure-freedom were 60.0% and 45.5% during the < 24-month and ≥ 24-month postoperative periods, respectively. The rates of seizure freedom for size ≥ 6 cm and < 6 cm were 25.0% and 58.6%, respectively. The rates of seizure freedom were 57.9% and 44.4% for patients with complete embolization and partial embolization, respectively. The rates of seizure freedom in arterial borderzone location and nonarterial borderzone location patients were 40.0% and 75.0% (p = .046), respectively (Table 2).

Table 2.

Seizure outcome in 37 AVM patients presented seizure after embolization treatment.

Characteristic	Modified Engel class I outcome	Modified Engel class II–IV outcome	p
Patients no.	19 (51.4)	18 (48.6)
Sex (male)	11 (52.4)	10 (47.6)	0.886
Age (mean ± SD)	30.7 ± 10.6	30.9 ± 15.8	0.973
Cerebral hemorrhage history
Yes	4 (66.7)	2(33.3)	0.709
No	15 (48.4)	16(51.6)
AVM location
Frontal-temporal	8 (44.4)	10 (55.6)	0.413
Other	11 (57.9)	8 (42.1)
Size
D < 6 cm	17 (58.6)	12 (41.4)	0.199
6 cm ≤ D	2 (25.0)	6 (75.0)
Eloquence	13 (50.0)	13 (50.0)	0.800
Nidus diffuseness Venous drainage patterns	7 (41.2)	10 (58.8)	0.254
Superficial drainage	17 (58.6)	12 (41.4)	0.199
Deep drainage	2 (25.0)	6 (75.0)
S-M grade, n (%)
I	3 (60.0)	2 (40.0)	0.337
II	6 (66.7)	3 (33.3)
III	7 (53.8)	6 (46.2)
IV	3 (37.5)	5 (62.5)
V	0 (0.00)	2 (100.0)
Side
Left	6 (37.5)	10 (62.5)	0.141
Right	13 (61.9)	8 (38.1)
Treatment modality
Embolization alone	8 (44.4)	10 (55.6)	0.413
Embolization and SRS	11 (57.9)	8 (42.1)
Obliteration rate
Complete	11 (57.9)	8 (42.1)	0.413
Partial	8 (44.4)	10 (55.6)
Arterial borderzone location
Yes	10 (40.0)	15 (60.0)	0.046
No	9 (75.0)	3 (25.0)
Seizure type
DRE	3 (37.5)	5 (62.5)	0.621
SPS	11 (57.9)	8 (42.1)
Other	5 (50.0)	5 (50.0)
Classification of epilepsy symptoms
Generalized seizures	12 (48.0)	13 (52.0)	0.637
simple partial seizures	7 (58.3)	5 (41.7)
Complex partial seizures	0 (0.0)	0 (0.0)
Postoperative time
≥24 months	10 (45.5)	12 (54.5)	0.385
<24 months	9 (60.0)	6 (40.0)
Taking AED status
Pre- and postoperative	3 (25.0)	9 (75.0)	0.012
No pre -and postoperative	3 (33.3)	6 (66.7)
Pre- and no postoperative	9 (81.8)	2 (18.2)
No pre- and no postoperative	4 (80.0)	1 (20.0)

AED: antiepileptic drug; AVM: arteriovenous malformation; DRE: drug-resistant epilepsy; SD: standard deviation; S-M: Spetzler-Martin; SPS: sporadic seizures.

New-onset seizure after Onyx embolization

One patient experienced new-onset seizures after embolization combined with SRS therapy. No patients experienced new-onset seizures after embolization alone.

AED use

In our cohort, 23 patients had ever taken AEDs before treatment, and 12 (52.2%) of them were still on AEDs. Nine of the 14 patients who were known not to be on an AED prior to treatment were receiving AED therapy. Among the 19 (51.4%) patients who reached modified Engel class I outcome, six (16.2%) patients took AEDs following treatment. Among the 18 (48.6%) patients who reached modified Engel class II–IV outcome, 15 (83.3%) took AEDs following treatment (Table 2).

Discussion

Seizures are the second most common clinical symptom in bAVM patients,¹ and uncontrolled seizures can considerably influence an individual’s mood and cognitive function. Surgical resection is always considered the gold standard for BAVM treatment.¹⁵ Previous studies always focused on seizure outcome after surgical and SRS, which may be because endovascular techniques are traditionally utilized as a supplement to surgery and SRS.

In this study, we found that seizure freedom was achieved in 19 (51.4%) of 37 patients who presented with epileptic seizure after embolization during 31.2 months (range, 12–62 months) of follow-up time. Of the 23 patients who had ever been treated with AEDs, 12 (52.2%) were still taking AEDs at last follow-up. One patient experienced a new-onset seizure. Arterial borderzone location was correlated with bad seizure outcome. bAVM hemorrhage history, frontal-temporal location and arterial borderzone location were associated with seizure occurrence before treatment.

Factors associated with pretreatment seizures

Although epileptogenesis in bAVM is not fully understood, potential proposed mechanisms include steal phenomena gliosis, demyelination, scar formation, repeated microbleeds with hemosiderosis, cerebral edema following venous obstruction, and kindling of excitatory synapses.^9,16–18 In our study, of the 239 patients evaluated, 68 (28.5%) presented with a history of seizures at intake. We found that several factors were associated with this initial history of seizures, including a history of bAVM hemorrhage, a frontal-temporal location and an arterial borderzone location. A recent prospective, single-center study found that 45 patients (29%) initially presented with a history of seizures, and that seizure occurrence was associated with male sex, frontal lobe bAVMs, arterial borderzone location and superficial drainage.³ Englot et al. found that hemorrhage, male sex and frontal-temporal location are associated with higher rates of preoperative seizures.⁶ Sex was found to be an independent factor associated with seizure in both these studies, though it was not related to seizure occurrence in our cohort. We found that a frontal-temporal location was an important factor in risk for seizure, in agreement with a previous study.⁶ We found that patients with seizures also had a higher rate of bAVMs in an arterial borderzone location. A previous study also reported that seizures at initial presentation were significantly associated (relative risk: 2.2) with bAVM arterial borderzone location.¹¹ This phenomenon may be that an arterial borderzone location includes cortical convexities, in line with an a prior report of a positive correlation between seizures and a cortical bAVM location.¹⁹

Seizure outcomes in patients with pretreatment seizure

We found that AVM-related seizure achieved an unsatisfied outcome in AVM patients with arterial borderzone location after embolization. On the one hand, we believe nonarterial borderzone-location bAVMs easily achieve higher obliteration rates and can relieve long-time steal phenomena gliosis that leads to seizure occurrence after embolization. At the same time, epileptogenic brain tissue can be rendered ischemic by embolization material in bAVMs with a nonarterial borderzone location, which may be associated with seizure-free status. Previous studies also indicated that complete BAVM obliteration achieved higher seizure remission rates after treatment.^2,4 On the other hand, seizure recurrence may occur in bAVMs located in an arterial borderzone location. In 2001, Ogilvy and colleagues reported that collaterals can develop rapidly after embolization, which reduced the effectiveness of the seizure treatment.²⁰ Studies by Sure et al. indicate that partially obliterated AVMs easily lead to transient regional hypoxia that can mediate neoangiogenesis within the bAVM nidus.²¹ We believe that the neoangiogenesis leads to recurrence of seizure. Jahan and colleagues also reported that neoangiogenesis and chronic inflammation can mediate the recurrence of seizure after embolization with Onyx in bAVM patients.²² Perinidal ischemic gliosis and residual gliotic plaques after embolization may also be the reason of poor seizure outcomes in patients with bAVMs.

It is important to note that there was one de novo seizure patient treated with embolization and SRS therapy in our study, but none who underwent embolization alone. A prior meta-analysis concluded that patients undergoing embolization alone had an elevated risk of de novo seizures.²³ De Los Reyes reported that four of 20 patients developed de novo seizures after Onyx embolization.⁹ We believe that hemodynamic change was not ineluctable for patients with bAVM after embolization, as this could have led to normal perfusion pressure breakthrough and perilesional edema,²⁴ which contributed to the occurrence of de novo seizures.

Limitations

There are some limitations to the present study. First, we retrospectively collected data from medical records in a single institution, and as treatment selection and referral patterns differ among institutions, the generalizability and reproducibility of our findings are limited. Additionally, electroencephalogram (EEG) evaluation (with conventional scalp EEG recording) was beneficial for the localization of an epileptiform focus; however, EEG was usually not used in our study. Moreover, as seizure status was evaluated by a telephone interview, these interviews were subject to some bias. For some patients, a shortened time after intervention restricted the window for analysis of postoperative seizure occurrence.

Conclusion

Patients with bAVM hemorrhage history, frontal-temporal location and arterial borderzone location were associated with seizures occurrence. Many patients treated with Onyx embolization still suffered an epileptic fit, but had been successfully weaned from antiepileptic medications. The classification of preoperative epilepsy was not associated with the seizure outcome. Patients with arterial borderzone location did not obtain seizure freedom but may have benefited in other ways. The seizure-free mechanism of bAVM with Onyx embolization is worth further study because AVM-related seizure is a serious hazard to patient quality of life.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: the National Natural Science Foundation of China (No. 81771233), and Specific Research Projects for Capital Health Development (2018-2-2041).