Angiographic characteristics of epileptogenic arteriovenous malformations and effectiveness in the seizure control after treatment with radiosurgery

Abstract

Objetive

Define the angiographic characteristics of epileptogenic arteriovenous malformations (AVM) and assess symptom control of the seizure after treatment with radiosurgery.

Material and Methods

Between 1996 and 2006, a total of 237 adults patients were diagnosed with AVM and were treated in our center by radiosurgery with linear accelerator. We analyzed demographics, clinicals, angiographics and radiosurgicals characteristics and the complications of the procedure in each of them. The first symptom was a seizure in 68 of them and the subsequent analysis of the treatment effectiveness for the seizure control was done, and the possible predictive factors of AVM nidus evolution were assessed.

Result

The average volume of the epileptogenic AVMs was 7.17 cc, compared ot the non-epileptogenic AVM 5.06 cc (p<0.03). Other differentiating factors were surface blood supply (p<0.003), venous ectasia (p<0.064), angiogenesis (p<0.078), and the presence of unrelated aneurysms (p<0.08). For 68 patients (28.7%) with seizures a clinical control (Seizure Frequency Scoring System SFSS ≤2) was obtained in 70% of patients and there was an excellent control (SFSS ≤1) in 25% of them. The percent occlusion of their AVMs was 50%. There was statistical significance with SFSS ≤1 (p<0.01), but was not any significance with SFSS ≤2. Age (p<0.003) and diffuse nidus morphology (p<0.05) were predictors of good AVM nidus evolution.

Conclusions

The stereotactic radiosurgery seems to be an effective method for control of symptomatic seizures for intracranial AVMs. Certain angiographic characteristics, such as the volume, surface blood supply, angiogenesis, venous ectasia, and unrelated aneurysms to the AVMs, seem to influence the appearance of epileptic seizures.

1. INTRODUCTION

The intracranial arteriovenous malformations (AVM) represent an important part of congenital vascular pathology, prevalent in approximately 0.14% of the population. There are different forms of presentation. They are, in order of frequency, intracranial hemorrhage, seizures, headache, incidental finding and neurological deficits attributable to hemodynamic steal syndromes [1]. The hemorrhagic presentation has been most studied because of the presentation’s greater frequency, morbidity, and mortality [2]. On the other hand, the literature is limited to the management of AVMs whose initial symptom is seizures, despite arising in one third of patients. Currently, the treatment of choice for an AVM is surgery whenever it is possible, as it immediately decreases teh chance of bleeding [3, 4]. However, there is a great controversy related to the indication of surgery for those AVMs associated with seizures, and only a few studies analyze the effect of radiosurgery on controlling them [5, 6].

The aim of our study is to observe whether the application of radiosurgery in patients with intracranial AVM, whose initial symptomatology was in the form of seizures, is effective in the clinical control of epileptic crises in the medium and long term. We further carried out this study to identify the characteristics of AVM angioarchitecture that may be related to cause seizure as the presentiang symptom, and to define whether there are clinical or angiographic outcomes to determine which patients with an AVM treated with radiosurgery will have good seizure control.

2. MATERIAL AND METHOD:

2.1. Patients and angiographic characteristics

Between 1996 and 2006, there were 237 adult patients who were diagnosed with intracranial AVM by arteriography and subsequently treated with linear accelerator radiosurgery (LINAC). Patient records were obtained from a retrospective database and the information was completed with medical records (hospital files, phone calls, correspondence, and iconography) and patient follow-up in the reviews of the medical consultation. We selected a total of 68 patients whose initial symptoms were seizures. Of these, 7 were excluded from the data analysis, because they did not complete at least one year of follow-up, and another one died before the first year.

For the follow-up data, the patients who suffered a first seizure and were subsequently diagnosed with AVM were collected, as well as those who had a continuous AVM nidus evolution. The seizures were classified as simple partial seizures, secondary generalized, complex partial seizures, and generalized tonic clonic seizures. We also collected demographic characteristics such as gender, age, and previous therapies. Each of the angiograms was evaluated by the team of interventional neuroradiologists of our center. The following angiographic characteristics were analyzed: diameter and volume of the AVMs, blood supply type, number of drainage veins and arterial feeders, angiogenesis, nidus morphology, presence of related aneurysms, venous stenosis, and ectasia.

2.2 Follow-up

A clinical follow-up was carried out in consultation and radiologically by cranial NMR at 3, 6, and 12 months and then every 6 month for three years, after which the patient undergoes to a control arteriogram. Patients with persistent malformative nidus were re-evaluated by the Radiosurgery Committee and a second treatment was proposed. These patients have been included when considering their evolution from the initial state.

The results in seizure control were considered a successful following the recommendations of the ILAE (International League Against Epilepsy), when patients had a seizure-free period of at least 12 months (including auras). The seizure-free period was defined and the need of antiepileptic medication was assessed, in order to classify patients according to the SFSS (Seizure Frequency Scoring System) modified by Engel [7](Table 1).

Table 1.

Engel- Seizure Frequency Scoring System

Seizure frecuency	Score
Seizure-free, not taking AEDs	0
Seizure-free, need for AEDs unknown	1
Seizure-free, requires AEDs to remain so	2
Non-disabling simple partial seizures only	3
Non-disabling nocturnal seizures only	4
Number of disabling seizures	5
1-3 per year	6
4-11 per year	7
1-3 per month	8
1-6 per week	9
4-10 per day	10
> 10 per day	11
Status epileptic	12

2.3 Radiosurgical technique

The radiosurgery was performed with a Varian linear accelerator of 6 MeV. At our center, we use a cranial fixation system with four points under local anesthesia, and the stereotactic guide Brown-Roberts-Wells was used. Planning was carried out using the software BrainLab, through the acquisition and fusion images of arteriography in two planes and CT. Circular collimators were used until 2003 when the use of micro-multilamina collimator was subsequently introduced. Planning was carried out by a multidisciplinary team of oncologists, physicists, radiologists and neurosurgeons.

2.4 Statistical analysis

For the statistical study, measures of central tendency and dispersion for quantitative variables (average, median, standard deviation, and percentiles) and the absolute and relative frequencies for the qualitative variables were calculated. The Chi-square test with Yates correction and Fisher’s exact test were used to study the variables. A p<0.05 value was considered significant. The statistical software SPSS 17.0 was used.

2.5 Ethics committee approval

This research was approved by Comité de Ética de la Investigación de Centro de Granada (CEI-Granada) of the Hospital Virgen de las Nieves (Granada).

3. RESULTS

We initially analyzed the clinical and angiographic characteristics in 237 adult patients treated at our institution, and we conducted a comparative study between those who had an epileptogenic AVM and who did not. Angiographic analysis showed that the average volume of the epileptogenic AVMs was 7.17 cc, significantly bigger that the non-epileptogenic AVMs, which average volume was 5.06 cc (p<0.03). The clear predominance of the superficial blood supply in epileptogenic AVMs when they are compared with the others AVMs was noted, with a strong statistical significance (p<0.003). Other features did not reach significance for the 95% confidence interval, but were close, in the cases of venous ectasia (p<0.064) and (p<0.078) angiogenesis in the epileptogenic AVMs. Likewise, there was a trend of association between AVMs and aneurysms distant from the malformative nidus, but without significance (p<0.08). No statistically significant differences regarding the nidus morphology, venous stenosis, the rest of arterial contributions, or with other related aneurysms was observed (Table 2).

Table 2.

Angiographic Characteristics

Characteristic	No seizure	Seizure	Significance
Nidus morphology; n (%)
Compact	87 (68,7)	35 (31,3)	NS
Diffuse	82 (68,6)	33 (31,4)	NS
Aneurysm; n (%)
None	134 (70,9)	55 (29,1)	NS
Intranidus	18 (78,3)	5 (21,7)	NS
Proximal	13 (86,6)	2 (13,3)	NS
Unrelated	4 (40)	6 (60)	0,086
Arterial supply; n (%)
Superficial	128 (67,7)	61 (32,3)	0,003
Mixed	24 (82,7)	5 (17,3)	NS
Deep	17 (89,5)	2 (10,5)	NS
Venous ectasia; n (%)
No	107 (75,9)	34 (24,1)	NS
Yes	62 (64,6)	34 (35,4)	0,064
Venous stenosis; n (%)
No	131 (72,8)	49 (27,2)	NS
Yes	38 (67,1)	19 (32,9)	NS
Angiogenesis; n(%)
No	148 (73,6)	53 (26,4)	NS
Yes	21 (58,4)	15 (41,6)	0,078
Volume, cc; mean	5,06	7,17	0,03

Abbreviation: n – total number, NS – no significance

Subsequently, those patients with epileptogenic AVM within the general series were analyzed and there were 68 (28.7%). We had to exclude eight patients who did not meet the minimum follow-up requirements, so the final number was 60. In 37 of these patients, the diagnosis of AVM was made after the first episode of seizure. The rest already had a known AVM, and had a variable period of attempted seizure control. The average age was 40.01 years (range 19-74 years), with a average follow-up of 46.72 months (range 2-144 months). The distribution by sex was 34 men (57%) and 26 women (43%). The type of main crises was recorded, which emphasized the generalized tonic-clonic crises in 25 patients (42%). Finally, the most common prior treatmnet before radiosurgery was embolization (16 patients, 23.5%) (Table 3).

Table 3.

Descriptive
Year old; (a) mean (range)	40,01(19-74)
Sex, n (%)
Male	34 (57)
Female	26 (43)
Type of seizure; n (%)
GTC	25 (42)
sGTC	9 (15)
Simple partial	8 (13)
Complex partial	2 (3)
N/A	16 (27)
Previous treatment; n (%)
Embolization	16 (23,52)
Surgery	6 (8,82)
Radiosurgery	4 (5,88)
Seizure debut; n (%)	37 (62)
Follow-up; (m) mean (range)	46,72 (2-144)

GTC: generalized tonic-clonic

The neuroradiology team analyzed the angiographic characteristics of the angiograms performed in patients with epileptogenic AVMs. The surface type blood supply was assessed, which predominated at the surface in 61 patients (89.7%). We also counted the number of drainage veins and feeders. The nidus morphology was studied, and a very similar distribution among compact and diffuse nidus was found (35 and 33 patients respectively). Also, the presence of aneurysms was noted (13 patients, 19.2%) – of these, the aneurysm was associated with AVM in 7 patients (10.4%), and independent each other in 6 patients (8.8%). The patients who presented angiogenesis (22%), venous ectasia (50%), and venous stenosis (28%) were noted. Finally, the diameters and volumes were determined, and we calculated the Spetzler-Martin grade of each of them (Table 4).

Table 4.

Angiographic Characteristics
Arterial blood supply; n (%)
Superficial	61 (89,7)
Mixed/Deep	7 (10,2)
Nº of draining veins; n (%)
≤ 2	52 (76,47)
>2	16 (23,52)
Nº of feeders; n (%)
≤ 2	16 (23,52)
>2	52 (76,47)
Angiogenesis; n (%)	15 (22,05)
Nidus morphology; n (%)
Compact	35 (51,47)
Diffuse	33 (48,52)
Aneurysm
None	55 (80,88)
Unrelated	6 (8,82)
Intranidal	5 (7,35)
Proximal	2 (2,94)
Venous ectasia; n (%)	34 (50)
Venosa stenosis; n (%)	19 (27,94)
Diameter; n (%)
<3 cm	43 (63,23)
3-6 cm	23 (33,82)
>6 cm	2 (2,94)
Volume; n (%)
>4 cc	45 (76,17)
<4 cc	23 (33,82)
Spetzler-Martin grade; n (%)
1-2	41 (60,29)
3-4	27 (39,70)

After the radiosurgery treatment, a follow-up was conducted in the consultation, registering a good control of seizures (Engel≤2) in 42 patients (70%). Of these patients, 15 (25%) had excellent control of seizures without drugs (Engel≤1). The average seizures management time was 53.28 months (range 0-126 months). Out of 60 patients, nidus obliteration was shown in 30 (50%), which did not show a statistically significant relationship with the degree of control of seizures (Engel≤2) (p<0.21). However, when the comparison was made with those patients with excellent seizures control (Engel≤1), a significant statistical relationship was found (p<0.01). A multivariate study of the different demographic and angiographic variables was made, which looked for a relationship with good control and an excellent control of seizures, and tried to determine some predictor of good response. The statistical significance was not reached except under two circumstances. Compact nidus morphology showed a relationship with a good control of seizures compared with diffuse morphology (p<0,043). On the other hand, patients under the age of 40 years were likely to present excellent control of seizures (p<0.003). Regardless of these factors, the radiosurgery with higher doses (equal to 16 Gy) and lower doses showed no statistical significance difference (p<0.064) (Table 5).

Table 5.

	Good seizure control (Engel=2)	Excelent seizure control (Engel≤1)	No seizure control	p
Patients; n (%)	27 (45)	15 (25)	18 (30)
Oclusion; n (%)	11 (36,6)	12 (40)	7 (23,3)	0,25a0,01b
Year old; a, n, (%)
< 40	4 (6,6)	11 (18,3)	8 (13,3)	0,52a
≥ 40	23 (38,3)	4 (6,6)	10 (16,6)	0,003b
Nidus morph; n (%)
Compac	17 (28,3)	10 (16,6)	5 (8,3)	0,05a
Diffuse	10 (16,6)	5 (8,3)	13 (21,6)	0,32b
Dose coberage;
< 16 Gy	5 (8,3)	2 (3,3)	7 (11,6)	0,064a
≥ 16 Gy	22 (36,6)	13 (21,6)	11 (18,3)	0,21b

Abbreviations: a, p evaluated for Engel ≤ 2; b, p evaluated for Engel ≤1.

Finally, the clinical and radiological complications were recorded at follow-up. There were 12 patients who presented clinical complications (19,9%), which were divided into four grades from minor to serious. There were ten minor cases of grade I (only headache) (16.6%), one patient with grade II (cutaneous disestesia minor neurological deficit), and another with grade III (decreased vision greater neurological deficit). There was no fatal complication (grade IV). Only one patient presented fatal hemorrhage during the follow-up, not related to radiosurgery. In terms of radiological complications there were two cases of radionecrosis, which were successfully conservatively treated. There was no secondary radiation-induced cancers. In addition, four patients had temporary alopecia after treatment (Table 6).

Table 6.

	Complications
Clinics; n (%)		Radiologics; n (%)
Grade I: headhache	10 (16,6)	Radionecrosis	2 (3,3)
Grade II: minus NL deficit		1 (1,6)
Grade III: greater NL deficit	1 (1,6)	Radioinduced tumor	0
Grade IV: exitus		0 (0)

Abbreviations: NL; neurological

5. DISCUSSION

Though the natural history of the AVMs in relation to bleeding risk is very well-documented, epilepsy control remains a controversial subject, despite this being the second-most frequent clinical presentation form. About one third of patients have it as their first clinical symptom [6]. However, there are no studies that clarify the predominant type of crisis for the AVMs. Schäuble et al. [8] noted the prevalence of simple partial seizures. In our case, generalized tonic-clonic seizures were predominant. Nonetheless, the results in terms of seizure control are similar.

The larger size of the AVMs has been identified as a risk factor for the development of seizures [6, 9-13]. The cortical localization is more related to the epileptogenic capacity [14]. Garcin et al. [13] relates to the epiliptogenic AVMs with the frontal lobe. Others, like Galletti et al. [15], observe the relationship with the temporal lobe. In our study, the cortical localization that is related with the surface blood supply, and the greater size of the AVMs, are related to an increased risk of seizures. In the majority of these peripheral AVMs, evidenced by superficial blood supply, the angiographic characteristics in relation to the debut in the form of seizure [12, 15] were noted. Yang et al. [16] identified a statistically significant relationship between the epileptogenic AVM volume and the degree in the classification of SFSS-Engel prior to radiosurgery. However, in our series, we have observed that there is a significant difference between the epileptogenic AVMs volume and the rest of the AVMs. This suggests that although we must take into account the effect of ionizing radiation on the brain parenchyma for seizure control, it is also important to consider that the size of these AVMs plays an important role in epileptogenesis. Sturiale et al. [12] recognized in their study the importance of topography and AVMs volume, although the malformative nidus size is considered most important. Similarly, Shankar et al. [14] attached importance to nidus morphology, watching for an increased risk of seizures with diffuse morphologies. In our case, although there were no differences between epileptogenic AVMs and all other AVMs, we have observed a relationship with good seizure control for compact nest morphology, which could be explained by the ease to apply radiosurgery on them. Other angiographic characteristics, such as the presence of venous ectasia, may be associated with epileptogenic AVMs. In large series of patients, the correlation between venous ectasia and development of seizures has been observed [12, 14, 17], although the etiology is not clearly defined. Furthermore there are no studies that have observed the relation between unrelated aneurysms and epileptogenic AVMs. On the other hand, there is still much debate about the role of angiogenesis in AVMs prognosis [18]. We have observed a trend, but not a significant difference, to the angiogenesis presence as morpho-structural angiographic characteristics, but we could not determine whether there was any connection with poor outcomes after radiosurgery.

The radiosurgery effect on the epileptogenics AVMs is unknown. However there is speculation that subcritical doses are sufficient to objectify a reduction in the number of crisis [19], thanks to the combined effect of ionizing radiation on angiogenesis [20] and the power imbalance that it produces [21]. This effect has classically been attributed to cell destruction that produces ionizing radiation. However, other neurophysiological, radiological, and histological studies raise questions about this assumption, as the tissue destruction has been shown as insufficient to explain the clinical effects obtained in almost all cases [22]. Therefore, the possibility that radiosurgery induces changes in the neural tissue functioning is increasingly accepted [23-25].

The beneficial effect of radiosurgery on the control of seizures induced by AVMs has been tested by several authors. Different series have obtained a good seizure control in between 26.6% and 80% of patients (see Table 7). Steiner et al. [26] has the largest follow-up series and his control percentage was 69%. Yang et al. [27] presents the longest series, with a control percentage of 76.6%. Our control percentage is 70% of patients, which is an acceptable number and also had long time follow-up.

Table 7.

Patients	Follow-up	Seizure control (%)
Heikkinen et al, 1989 [9]	29	2-6 a	16 (55)
Steiner et al, 1992 [27]	59	24-96 m	41 (69)
Gerszten et al, 1996 [29]	15	47 m	11 (73)
Kurita et al, 1998 [30]	35	43 m	28 (80)
Schäuble et al, 2004 [8]	65	4 a	48 (74)
Lim et al. 2006 [31]	43	46 m	23 (53,3)
Yang et al. 2012 [28]	86	92,5 m	66 (76,6)
Wang et al. 2013 [32]	30	38,1 m	8 (26,6)
Cordero et al. 2013	60	46,7 m	42 (70)

Schäuble et al. [8], under the light of the results of their latest study, achieved greater long-term seizure control when the injury treated has a smaller size and lower volume. However, Steiner et al. [26] did not find a significant relationship between these facts, and noted that a radiological control of the injury is not necessary to observe clinical improvement. This has been documented by other authors, who did not see a relationship between occlusion and seizure control [8, 9, 26]. On the other hand, several recent studies have already observed this relationship, seeing that those patients with AVM obliteration had better results in controlling long-term crisis [16, 27-30]. In our study, we observed only the relationship between occlusion and positive outcomes in patients who had excellent seizure control (Engel ≤ 1), so we believe that, in fact, a good or bad outcome depends on several factor. Whether it is logical or not to think that a complete obliteration should solve the problem, many patients without complete obliteration have good seizure control, so the effect of radiation should also have a significant impact.

Many authors have found that surgery in the AVM treatment allows for good control of epilepsy [31-34]. With our results, radiosurgery as a therapeutic technique is postulated valid because it achieves results comparable to published surgical series. However, one limitation is the increased bleeding risk of latency time.

Our study conclusions can be taken when conducting future research, as it would be interesting to evaluate the radiation neuromodulation. Likewise, angiographic characteristics, such as nidus morphology or venous ectasia, seem established as potential prognostic factors. Future research still needs to define their pathophysiological mechanisms.

6. CONCLUSION

The epileptogenic intracranial AVM treatment using radiosurgery has shown effective not only for nidus malformation occlusion and protection from bleeding, but also for the clinical assessment of associated seizures. Its action mechanisms are not fully understood, as there is no clear direct relationship with the occlusion degree. Certain AVM angiographic characteristics, such as volumen or superficial blood supply, seem associated with an increased risk of seizures. At present, we lack clinical guidelines for AVM management, making it necessary to perform prospective randomized studies in order to establish the appropriate treatment indication of epileptogenic AVMs.