31 Epilepsy Management for Arteriovenous Malformation
About one-third of patients with arteriovenous malformation (AVM) present with seizures, and about 40% of those will have AVM-related epilepsy (AVMRE). Seizures will persist despite best antiepileptic drug (AED) therapy in about 15 to 20% of those with AVMRE. Risk factors for development of AVMRE described by most studies include superficial cortical location–usually in frontotemporal regions–and previous hemorrhage. Intervention for AVM is primarily undertaken for risks of hemorrhage, but surgery, embolization, or stereotactic radiosurgery (SRS) have all demonstrated reasonably high rates of sustained seizure remission in those with AVMRE. Comparative studies suggest that open surgery may have more beneficial effects in seizure remission than other treatment modalities in patients with AVMRE. Conversely, SRS has fewer problems associated with de novo, postoperative AVMRE than surgery in those patients with AVM and without preoperative seizures. Modern practice with multiple modalities with embolization, SRS, and open surgery in a staged fashion has demonstrated efficacy in seizure remission that correlates with obliteration of the AVM nidus, regardless of intervention.
Keywords: epilepsy, anticonvulsant medication, seizure remission, arteriovenous malformation, sudden unexpected death in epilepsy, subarachnoid hemorrhage, radiosurgery, embolization
- About one-third of patients with arteriovenous malformation (AVM) present with seizures, and about 40% of those will have AVM-related epilepsy (AVMRE).
- Seizures will persist despite best antiepileptic drug therapy in about 15 to 20% of those with AVMRE.
- Risk factors for development of AVMRE described by most studies include superficial cortical location–usually in frontotemporal regions–and previous hemorrhage.
- Surgery, embolization, or stereotactic radiosurgery and combinations of interventions have all demonstrated reasonably high rates of sustained seizure remission in those with AVMRE.
- Efficacy in seizure remission correlates with obliteration of the AVM nidus, regardless of the mode of intervention.
Many patients with arteriovenous malformation (AVM) present with seizures and can be considered as having epilepsy; an important minority remains with recurring seizures despite maximal therapy. Although intracranial hemorrhage and its attendant morbidity and mortality remain appropriate foci of study of AVM, epilepsy adds to the risks of AVM and merits additional consideration in the overall care of the patient with AVM-related epilepsy (AVMRE).
The prevalence, characteristics, and responses to treatment of epilepsy associated with AVM (AVMRE) will be outlined placing this specific disorder in the context of epilepsy in general. The mechanisms of epileptogenesis will also be briefly discussed.
31.2 Definitions, Morbidity, and Mortality of Epilepsy
A practical definition of epilepsy is “at least two unprovoked seizures occurring more than 24 hours apart” or “one unprovoked seizure and a [high] probability of further seizures.”1 Seizures are considered “medically intractable” when a patient has failed to become seizure free with trials of at least two anticonvulsant medications (antiepileptic drugs [AED]).2 Rather than layering on two or more AED, most epileptologists recommend serial trials of single AED since side effects and teratogenicity of AED rise rapidly, and seizures may paradoxically worsen, with polypharmacy.3
A recent review by Laxer et al lays out the epidemiology and risks of epilepsy.4 The incidence of epilepsy in developed countries is approximately 50 per 100,000 individuals per year. The prevalence of epilepsy in developed countries ranges between 4 and 10 per 1,000 individuals per year. About 20 to 30% of patients with epilepsy can be expected to be “medically intractable,” and only 4% of adult patients can expect to remit per year even taking into account all current therapies.4 About 50% of patients can be expected to achieve sustained seizure freedom with the introduction of a single, well-tolerated AED; a trial of a second AED brings that proportion to about 65%.5
The most common scheme for rating seizure outcomes after epilepsy surgery is the “Engels criteria” that classify outcome 2 years after surgery into four grades and subgrades: Ia: seizure-free; Ib: free from all disabling seizures (effectively no seizures except for simple partial seizures or auras); II: rare disabling seizures; III: worthwhile improvement, and IV: no worthwhile improvement.6 Quality of life (QOL) or functional outcomes after “lesionless” epilepsy surgery show meaningful or durable improvements only in the case of class I outcomes.7,8,9 In other words, improvements in epilepsy do not occur linearly; a 50% improvement in seizure frequency, for example, does not translate into a 50% improvement in cognition, driving, or employment. Most benefits in epilepsy treatment accrue to those who are truly seizure free (Engels Class I) rather than to those whose seizures have merely improved (Engels Class II or worse).
Accordingly, burdens of intractable epilepsy are considerable. In a comparative study from 2011, the global burden of epilepsy for women was greater than that of breast cancer, and was nearly four times greater than that of prostate cancer for men.10 Because this comparison omits epilepsy’s effects of stigma and social exclusion, the actual burdens may be even higher. People with epilepsy, in comparison to matched U.S. Census bureau norms, receive less education, are less likely to be married, employed, or driving, and are more likely to have diverse psychosocial problems and subjectively worse lifestyles.11 Dysfertility, sexual dysfunction, and other consequences of hormonal and sleep dysregulation worsen overall QOL.12
Many studies of AVM center on expected rates of death or major disability from hemorrhage. Epilepsy itself, in addition, carries an additional risk of early death. Mortality is greater for those with epilepsy than for those without for many reasons, including sudden unexpected death in epilepsy (SUDEP), accidents, suicide, vascular disease, pneumonia, and factors directly related to underlying causes (e.g., AVM, but also including brain tumors, neurodegenerative disease, and cerebral vascular disease, among others). Overall, people with epilepsy have a 1.6 to 11.4 times greater mortality rate than expected.4,13 Many centers now focus on SUDEP, especially given its higher-than-expected incidence in placebo-controlled arms in standard randomized, controlled AED trials.14 The average incidence of SUDEP is 1 per 1,000 patients with epilepsy per year. SUDEP is more common in those with medically intractable epilepsy. In adults with refractory epilepsy, the incidence is 6 per 1,000 patients per year, and the lifetime incidence is 7 to 35%.15
Therefore, epilepsy is a common and important disorder with psychiatric, psychological, and medical consequences. The information available on AVMRE should be evaluated in this context.
31.3 Prevalence, Incidence, and Characteristics of AVMRE
One shortcoming that quickly becomes evident when reviewing the literature on the intersections between AVM and seizures is that authors, especially in earlier studies, often fail to explicitly distinguish among “presenting seizures,” recurring seizures, or medically intractable epilepsy.
31.3.1 Seizures as Presenting Sign in Arteriovenous Malformation
Some studies focus on the initial presenting sign or symptom of AVMs as opposed to other possible symptoms or signs. For example, a large multicenter study by Hofmeister et al gathered 1,289 patients with AVM from three centers and evaluated the broad demographics of AVM.16 The initial presenting symptoms were, in order, hemorrhage (54%), seizures (40%), chronic headache (14%), and focal neurological deficits of varying time courses (persisting, progressive, or transient, totaling to 20%). Some studies of AVMRE excluded patients with hemorrhage. In the studies of unruptured AVM, Al-Shahi Salman et al17 and Garcin et al,18 the largest group of patients, about 50%, presented with “incidental” AVMs. Initial seizures comprised 29 to 41% of presenting signs.
31.3.2 Prevalence of AVM-Related Epilepsy
► Table 31.1 summarizes the results of studies that report the occurrence of AVMRE beyond the initial presentation; most of these studies do not clearly distinguish between presenting seizures and preoperative continuing seizures. Given that limitation, the average prevalence of AVMRE before surgical intervention is 32% (range: 18–57%).
Table 31.1 Incidence of “presenting seizures” and epilepsy in studies of arteriovenous malformation–related epilepsy
Estimates of the severity and medical intractability of AVMRE are difficult because of variations in definitions and study designs. Three studies focused on presurgery seizures with definitions clearly stated regarding medical intractability. Osipov et al described the characteristics of seizures before surgical treatment of unruptured AVMs in a prospective cohort of 328 patients, of which 92 (28%) presented with seizures unrelated to hemorrhage. Of these 92 patients, 62% of this sample had only a single seizure at presentation; 11% continued to have weekly, 19% monthly, and 9% annual seizures. Since all patients were treated with AED, the incidence of medically intractable AVMRE was 35% of the sample of those who presented with seizures and 11% of the total sample who presented with unruptured AVMs.
Englot et al defined preoperative and postoperative outcomes in terms of “medical intractability” using the rigorous definition of failure of two AED trials.19 Out of 440 patients with AVM, a total of 130 (30%) experienced preoperative seizures. Of these individuals, 98 (75%) had seizures as their presenting symptom, and 23 (18%) progressed to medically refractory AVMRE. It should be noted that this estimate derived from a rigorous definition falls under the average prevalence obtained from ► Table 31.1 and occupies the lower range commonly accepted as medically intractable epilepsy in general.
A prospective study by Josephson et al evaluated the longitudinal risk of development of epilepsy after presenting with a first AVM-related seizure in a sample that included those patients who experienced hemorrhage as well as those with unruptured AVMs.33 The 5-year risk of developing epilepsy was 58% (95th confidence interval: 40–76%). The proportion of patients with unruptured AVMs who achieved a 2-year span of seizure freedom was 45% of those with epilepsy. Therefore, in this sample of unruptured AVMs, 55% remained with continuing seizures.
Some studies define the severity of seizures as “disabling” or “severe” seizures, based on frequency, intractability, or type without providing clear definitions. This is not a judgment against authors: “disabling” in Engel’s criteria is stated without definition as well.6 To paraphrase Potter Stewart in his Supreme Court concurrence, disability, like pornography, may be hard to define, but we know it when we see it. Many physicians equate the loss or alteration of consciousness as the primary consideration in defining disabling (i.e., as experienced with complex partial seizures). Although simple partial seizures spare consciousness, their interruptive and anxiety-provoking aspects may disproportionately affect QOL, an opinion supported by evaluations of QOL by seizure type.34
The types of seizures that occur with AVMs should, theoretically, be confined to focal seizures with or without secondary generalization, presuming that an AVM causes a discrete epileptic focus, or at most a limbus of adjacent epileptic tissue. Since secondarily generalized seizures can appear to witnesses as primary generalized seizures (especially those that propagate rapidly from the focus), listings of seizure types unsupported by electroencephalographic (EEG) findings should be viewed cautiously. Many reports feature a surprisingly high distribution of “generalized tonic–clonic” seizures with an across-study average of 63% (range: 5–81%; ► Table 31.2). In our subspecialty epilepsy clinic, simple partial seizures both before and after intervention are the predominant seizure type in AVMRE, an experience reflected by the proportions reported by a most recent study20 as well as one of the more thorough early studies.21 The across-study average of simple partial seizures is 25% of all seizure types (► Table 31.2).
Table 31.2 Distribution in seizure types reported in arteriovenous malformation–related epilepsy
31.3.4 EEG Assessment of Preoperative AVM-Related Epilepsy
EEG is a commonly used tool in assessment of epilepsy, but few studies of AVMRE report EEG findings, and only one appears to have evaluated EEGs uniformly across their samples. In a case series in which all patients underwent epilepsy surgery for intractable AVMRE and in which all underwent routine EEG recorded from the scalp, 37% of patients had interictal epileptiform discharges37 (IEDs; the principal abnormality predictive of epilepsy present in patients with epilepsy in between seizures).
Other studies report the incidence of IEDs of those for whom EEG results were available. A recent case series of AVMRE documented that 13% had focal interictal epileptiform discharges.20 Earlier studies show sensitivity of routine EEG between 38 and 86%.38,39 In comparison, the best estimates of sensitivity of a single, routine EEG for patients with known epilepsy of any cause range between 29 and 55%.40,41,42 Therefore, the rate of IEDs found preoperatively in AVMRE is within ranges reported for epilepsy in general.
Long-term video EEG, as opposed to routine EEG, is used to localize the epileptogenic zone before epilepsy surgery by capturing seizures rather than just IED. However, no studies are available to describe findings with this standard preoperative technique. The closest report with uniform EEG use was Yeh et al who used a combination of preoperative chronic EEG to capture “EEG abnormalities” and intraoperative electrocorticography (ECoG) to evaluate the location of the epileptic focus in relation to AVM.43 They documented that in 46% of subjects, with the use of these combined techniques, an epileptic focus was found. Although generalized seizures were not seen, in 44% of those with IEDs, remote foci in the mesial temporal structures–supposedly regions of secondary epileptogenesis–were seen beyond the regions of AVM.
31.3.5 Risk Factors of AVM-Related Epilepsy
Demographic or AVM characteristics that may place patients at higher risk of presenting with seizures or continuing with AVMRE vary among reports (► Table 31.3).18,19,20,21,22,23,24,33,35,44,45 Regarding demographics, a minority of studies found that male sex18,19,24 and younger age24 carried higher risks for AVMRE. Two studies contradicted each other over prior AVM surgery.21,24 Ding et al noted that prior AVM embolization worsened the likelihood of AVMRE.45 Patients with neurological deficits associated with their AVM were found to have higher rates of AVMRE in two studies.22,46
The possible contribution of hemorrhage to the development of AVMRE was nearly uniformly evaluated in those studies that included patients with ruptured AVM. The majority of studies found that prior hemorrhage of AVM contributed to the likelihood of developing or presenting with AVMRE.19,21,46 By virtue of its prospective acquisition and careful longitudinal followup, the study by Josephson et al may be considered the most rigorous.33 The 5-year risk of first-ever seizure after presentation was higher for AVMs presenting with intracranial hemorrhage or focal neurologic deficit (23%) than for incidental AVMs (8%). An early study of a cohort of 343 patients followed longitudinally similarly found that development of epilepsy was confined to those with hemorrhage at a rate of 1% per year in absence of surgical treatment.21 Englot et al, more recently, found that 43% of those with prior hemorrhage had AVMRE as opposed to 13% without hemorrhage.19 Counter to the above studies, a retrospective study by Ding et al found that those patients without prior hemorrhage before undergoing SRS were 22 times more likely to have preoperative AVMRE than those with hemorrhage.45
Studies differ in the effects of AVM characteristics in association with epilepsy. Studies split almost evenly finding that larger AVMs have higher rates of AVMRE. Notably, the prospective incidence study by Josephson et al found that AVM size had no significant risks for AVMRE.46 The location of AVM is a more consistent predictor of AVMRE; most studies determined that frontal, temporal, or frontotemporal locations were particularly susceptible to AVMRE.18,19,20,23,24,44,46 Supporting these findings were Galletti et al and Ding et al who found, in a complementary fashion, that AVMs in the occipital region were particularly less susceptible to AVMRE.20,48 Other facets of AVM morphology such as pseudoaneurysms, a single draining vein, superficial versus deep venous drainage, and Speltzer-Martin grade carried mixed associations with AVMRE.
31.3.6 Morbidity and Mortality of Preoperative AVMRE
Few studies have evaluated preoperative morbidity or mortality of AVMRE as a factor distinct from that of AVM in general. Regarding perceived morbidity, one study of SRS-treated patients surveyed them early in the latency period after SRS but before the emergence of radiosurgical or clinical changes. “Irreversible physical disabilities” had the most unfavorable effect on QOL compared to “reversible” symptoms such as epilepsy, headache, or transient physical symptoms.49
Prospective studies on the natural history of AVM may have data pertinent to seizure-related mortality hidden within causes of death. For example, the comparison of intervention versus conservative therapy for AVM by Al-Shahi Salman et al found that the overall risk of mortality over a 12-year follow-up period due to AVM was higher after intervention than for conservative management.17 Forty-one deaths occurred across both arms; the cause of death was listed as “other” for 26 (62%) compared to an AVM- or intervention-related death. Certainly, SUDEP, seizure-associated trauma, or other complications of epilepsy could lay hidden in “other” deaths in this cohort.
31.4 Outcomes of Treatment for AVM-Related Epilepsy
AVMRE is merely one reason among several–future hemorrhage, worsening neurological deficits, etc.–to proceed with interventions. Several prospective comparisons of intervention versus conservative management, patients–who are not randomized–usually are younger and more likely to have AVMRE than patients followed in the conservative arms.17,46 Thus, patients or their physicians may feel that AVMRE is an important factor in undertaking more aggressive therapy.
31.4.1 Outcomes of Open Surgery
► Table 31.4 contains the percentages of patients who achieve EC1 seizure remission after open surgery for AVM across studies published between 1972 and 2015. Although mean follow-up durations after surgery are sufficiently long, some studies include only minimal follow-up far shorter than the 2 years specified in the Engel’s classification. With that limitation in mind, seizure remission rates are between 4 and 93% with a trend of improvement over time. A close look at the earliest study in ► Table 31.4, Forster et al, gives a snapshot of outcomes regarding epilepsy before the widespread use of presurgical embolization and microsurgical techniques.50 In this study, patients with AVM who presented with either hemorrhage or epilepsy were followed long term (mean > 15 years) after partial or total removal of AVMs. Of the 104 patients with AVMRE, only 4% achieved complete seizure remission after surgery. Subanalyses to evaluate factors in favor of seizure remission were not evaluated. Later studies that presumably benefit from modern techniques show marked improvements in seizures outcomes over the 40-year span represented in ► Table 31.4. Englot et al established in their cohort that EC1 outcome was achieved by 117 of the 126 patients (93%) who had preoperative seizures.19
Table 31.4 Summary of surgery series for arteriovenous malformation–related epilepsy