Seizures in Patients with Brain Metastases: A Literature Review

The number of patients living with brain metastases (BMs) has increased significantly over the last 5-10 years. This is due to a combination of improved survival resulting from better systemic treatment options for many of the primary cancers most commonly associated with BM, and better treatment of intracranial disease. In addition, the increased availability of magnetic resonance imaging (MRI) and better quality of imaging has resulted in more asymptomatic patients being diagnosed with small volume intracranial disease.


Introduction
The number of patients living with brain metastases (BMs) has increased significantly over the last 5-10 years. This is due to a combination of improved survival resulting from better systemic treatment options for many of the primary cancers most commonly associated with BM, and better treatment of intracranial disease.
In addition, the increased availability of magnetic resonance imaging (MRI) and better quality of imaging has resulted in more asymptomatic patients being diagnosed with small volume intracranial disease.
Many patients with BM have a limited prognosis of <6 months and are treated with best supportive care with or without whole brain radiotherapy. However, an increasing number of fit patients with controlled, controllable or absent systemic disease are being referred for surgical resection and/or stereotactic radiosurgery (SRS) or fractionated stereotactic radiotherapy (FSRT) for BM.
The median survival for these patients is 12-15 months, but with improved control of extracranial disease in many subtypes of cancer, the population of patients with BMs is recognised as a heterogeneous group with some patient subgroups living for years with BM, and consideration of the longer-term side effects of treatments is increasingly important.
Seizures are one of the most common presenting symptoms of BM [1][2][3][4][5] and may be the presenting symptom of intracranial disease or even of the underlying cancer. A further cohort of patients develops de novo seizures following treatment [6].
However, much of the seizure incidence data comes from patients treated prior to the availability of newer systemic treatments such as immunotherapy and targeted therapies. The ban on driving is usually upheld for at least 1-2 years and this driving ban is the single most difficult issue for patients with BM in clinic. The majority of patients do not have seizures at presentation nor following treatment and, with the increasing inclusion of cranial imaging in staging asymptomatic patients, the proportion of patients with BM who experience seizures is likely to be significantly smaller than previous estimates.
In this paper, we review published literature on seizure incidence in patients with BMs and discuss the possible pathogenesis of seizures in this cohort in comparison with primary brain tumourassociated seizures. We also present the literature discussing risk factors for development of seizures before and after treatment.
Although only 5% of patients with extra-cranial malignancy develop seizures, estimates in the population with BM range from 10-40% [2,5,6,8,11,12] similar to the incidence reported in patients with primary brain tumours [13]. Higher rates of up to 67% have been reported in patients with BM from melanoma [4].

Pathophysiology of Seizure Development with Different Tumours
The pathophysiology of seizure development ('epileptogenesis') in patients with primary or metastatic brain tumours is not fully understood but is likely to be multifactorial. Different pathogenic processes may play a role in seizure development for BM from different primary tumour types, and for those developing seizures at diagnosis versus post-treatment. These differences may explain the variation in seizure frequency between primary tumour types [14][15][16].
Tumours such as intrinsic glioneuronal tumours (e.g. gangliogliomas and dysembryoplastic neuroepithelial tumours, DNTs) exhibit intrinsic epileptogenesis which is believed to arise from overexpression of neurotransmitters and neuropeptides causing a hyper-excitable neuronal component and disruption of the normal balance between inhibitory and excitatory stimuli [17].
The disorganisation of the adjacent cortex may also contribute to seizure generation.
For slower-growing low-grade primary brain tumours, which cause seizures in approximately 60-88% of patients [18,19], seizure activity may be caused by hypersensitivity of surrounding brain tissue [20][21][22]. Disruption to the balance of the normal neural networks is thought to contribute to the epileptogenesis with the formation of 'random networks' with lower seizure thresholds [15,23]. Epileptogenic foci may also be found distal to the primary tumour and 'secondary epileptogenesis' may arise from alterations in peritumoral tissue [20].
In de novo high grade primary brain tumours such as glioblastoma multiforme, the risk of seizures is lower than for low-grade tumours at 30-50% [24,25]. The mechanism of seizure development is thought to be through the abrupt tissue damage with necrosis, oedema, bleeding and haemosiderin deposition.
In the peri-tumour microenvironment, reduction of glutamate synthetase may lead to impaired metabolism of the excitatory neurotransmitter glutamate to glutamine, resulting in enhanced excitation of surrounding neural networks [26]. The additional burden from tumour growth and aberrant angiogenesis could lead to hypoxia, acidosis, free radical generation and downstream damage to alterations in gene expression, unstable cell turnover and seizures. In extrinsic tumours such as meningiomas, the mechanism of epilepsy development is thought to be due to peritumoral oedema [27].
In BM, the pathogenesis is most likely to be similar to that of high-grade gliomas with the hyper-excitability state arising from the effects of the tumour on the adjacent neural tissue, although the precise mechanism is poorly understood and a number of factors may contribute to the risk. Extrapolating possible mechanisms from the evidence from primary brain tumours, there may be altered expression of neurotransmitters and receptors in peritumoral tissue and high levels of the excitatory neurotransmitter glutamate has been found in the brain tumours of patients with epilepsy. Activation of glutamate receptors may down-regulate γ-aminobutyric acid (GABA)-mediated inhibitory stimuli which may contribute to tumour-related seizures. The difference in seizure incidence between BM subtypes and sites is not understood and causative factors remain unknown. and examined risk factors in both cohorts. In the SEER-Medicare cohort, Black race, urban residence, melanoma and stereotactic brain radiation were associated with greater seizure risk. In the institutional cohort, melanoma, more than 4 BM, BM in a high-risk location and lack of brain-directed therapy were important risk factors and the authors suggest prophylactic anti-epileptic drugs (AEDs) be considered in patients with melanoma metastases, a large intracranial disease burden or metastases in high-risk locations [11]. In a separate retrospective study of 565 patients undergoing surgical resection, 20% presented with seizures at diagnosis. In this population, headache, cognitive decline, the number of BM (≥ 2), tumour location (temporal lobe and occipital lobe), and bone involvement were associated with increased seizure risk at presentation [2].
Results from other reviews show a variation in seizure incidence with primary tumour histology with the highest incidence in those with melanoma, followed by lung cancer and gastrointestinal cancer and breast cancer, although results vary between series' [4,8] and this may reflect changes in imaging and treatments available over the years. As well as being at higher risk of developing multiple BMs, itself a possible risk factor for seizure development, patients with melanoma and lung cancer are considered at higher risk of seizure due to the increased risk of haemorrhage, independently thought to be associated with seizure. However, much of the data is from patients treated prior to the availability of immunotherapy and the relevance of the data to today's patients is questionable.
As mentioned above, tumour location may also affect seizurerisk but this too has not been demonstrated consistently between studies, with some suggesting the frontal lobe as having the highest risk, whilst others report the temporal lobe, insula and/or occipital lobe as being important for seizure-development [1,2,28].
The number of BM may also be important with increased seizure incidence reported in patients with multiple BM in some retrospective reviews [2,29], although here too results are inconsistent with some studies reporting increased incidence with a solitary metastasis1. In patients with 1-3 BMs, the prevalence of epilepsy prior to SRS or FRSRT was 33/195 (16.9%) with a significantly higher risk in patients ≤61years old, with trends (although not significant) in those with primary NSCLC and controlled extracranial disease [30]. Interestingly, symptoms and seizures prior to radiotherapy do not appear to be related to survival.
Other factors, not directly related to the presence of BMs may influence seizure activity, including systemic treatment, metabolic disturbance, opportunistic infection and paraneoplastic encephalitis. The number of cases of status epilepticus was found to have increased significantly in one single-centre retrospective study since the introduction of immune checkpoint inhibitors (ICIs) in 2014. However, the number of cases was small and further larger prospective studies are required [31].

Seizure Activity Post-Radiosurgery
It has been suggested that radiotherapy, whether whole brain, partial brain or SRS, may provoke seizure activity by increasing brain oedema (acute adverse event) or as a manifestation of radiation necrosis (late toxicity). A few limited retrospective studies have investigated the relationship between BM, its treatment, and the development of seizures but much of the literature also includes patients with primary brain tumours, and in particular low grade glioma [13][14][15]28]. In a separate retrospective study, 32/258 (12.4%) patients with lung cancer BMs, treated with gamma knife radiosurgery developed de novo seizure activity after treatment [33].
The reported incidence of radionecrosis in a cohort of 206 patients with single or multiple BMs treated with SRS was 24% (75/310 treated lesions), with symptomatic radio-necrosis reported in 10%. Reported symptoms were: seizure (5.2%, 16 patients), motor deficits (2.9%), cognitive deficits (1.0%), sensory deficits (1.3%) and speech deficits (1.3%). Three patients developed epilepsy without radionecrosis. The authors suggest hypofractionated stereotactic radiotherapy may be a better option for patients if the normal brain volume receiving 12Gy (V12Gy) was greater than 8.5cc, as the risk of radio-necrosis was nearly 10% [34].

Seizure Activity Post-Craniotomy
The incidence of seizure post-craniotomy is thought to be 15-20% with most events occurring within the first month after surgery. Although prophylactic AEDs may be used temporarily in the immediate peri-operative period, there is no evidence to support their use in reducing the risk of seizures at other times.
In the series by Wolpert discussed above, the incidence of seizures post-operatively was 15.4% [1]. Factors associated with post-operative seizures were supratentorial disease (p=0.012), occipital location (p=0.027), venous thrombosis (p=0.03) and single versus more than one brain surgery in the disease course (p=0.00001). Gross total resection was favourable but brain irradiation increased seizure risk (p=0.013), as did chemotherapy (p=0.002). Interestingly there was no association between preoperative and post-operative seizures (p=0.077). In contrast to pre-operative findings, tumour histology was not associated with post-operative seizures (p=0.479). On mulitvariate analysis, supratentorial disease and subtotal resection were independently associated with post-operative seizures, although the risk was not statistically significant for multiple surgeries (p=0.095).
In the series by Wu described above, post-operative seizures occurred in 5.3%, but here the risk of seizures was increased by pre-operative seizures, temporal lobe location, absence of systemic therapy after surgery, subtotal resection and local relapse [2].

Seizure Treatment and Prevention
Monotherapy with AEDs is recommended for those patients with brain tumours who have had at least one seizure event. The preferred first line AED is Levetiracetam given its tolerability, effectiveness and lack of drug interactions [35]. A prospective observational trial evaluated the safety and efficacy of levetiracetam, oxcarbazepine and topiramate as monotherapy in 48 patients with brain metastases and found that 63% of patients remained seizure free after a mean follow up of 6 months [36].
Approximately 50% of people with tumour-related epilepsy may respond to monotherapy with AEDs [37]. Retrospective analysis of the risk of seizures in patients with melanoma BMs from a single centre showed that seizure led to the diagnosis of BM in 13% (14/109) and a further 20% (22/109) developed seizures later. In those without a seizure at diagnosis prophylactic AEDs significantly decreased the risk of seizure with 3-month seizure rate of 0% compared to 17% without prophylactic AEDs.
Univariate analysis demonstrated that haemorrhage and multiple supratentorial metastases were associated with increased seizure risk and the authors suggest that prophylaxis may be beneficial in selected patients, and this should be addressed in a randomized controlled trial [38].

Guidelines recently published by the Congress of Neurological
Surgeons and endorsed by the American Society of Clinical oncology (ASCO) and Society for Neuro-Oncology (SNO) recommend that prophylactic AEDs are not used routinely in patients with BMs who did not undergo surgical resection and are seizure free, nor for those post-craniotomy who remain seizure-free except for a short period of use perioperatively for seizure control [39][40][41][42][43]. However, as discussed by Goldlust et al. [38] the data available when designing these guidelines included heterogenous patient populations and among the 4 randomised controlled trials, only one considered BM as a distinct cohort from primary brain tumours, and the systemic management will be outdated [38,[44][45][46][47].

Conclusion
At the time of diagnosis of BM, all patients in the UK are informed that they are no longer allowed to drive because of an increased risk of developing a seizure. In many other countries the treating doctor decides whether an individual with BM who is currently seizure-free should continue to drive or not, but there are few data to support specific patient-tailored and evidence-based To date there is no evidence to support the use of prophylactic anti-epileptic drugs in patients with BM. However, here too the current data available are based on limited data; and there may be subgroups of patients with BM who would benefit from AEDs.
If those at high seizure risk can be identified, further trials can specifically assess the benefit of AEDs in this population, and thereby potentially reduce the seizure rate in those most at risk. Furthermore, whilst it is imperative that public and individual patient safety is paramount, the current driving guidelines in the UK may be stricter than required for safe practice. The threshold seizure risk that is usually set for patients with Group 1 licences i.e. non-professional drivers is 20% per annum and therefore any study that can demonstrate a lower risk than this will potentially advise the next guidelines so that a larger number of patients with BM can continue driving, with a significant improvement in their quality of life.