Epilepsy Treatment: A Futurist View

Abstract

Advances in epilepsy treatment are occurring at a rapid pace, and it is challenging for us to keep up with the latest in our field. As we struggle to keep up with the literature and concentrate on our own research and clinical work, we often fail to exercise our imagination and envision what our field will be like in future decades. This was the assignment to the speakers for the Presidential Symposium at the 2016 American Epilepsy Society Annual Meeting. I challenged the experts to step outside the frame of their usual daily work to imagine what epilepsy treatment would and should look like for the next generation of epilepsy specialists and their patients.

As you will read in the following sections, the speakers truly stepped up to the challenge to look into the crystal ball. The following are summaries of each lecture that describe the current state, existing cutting edge ideas, and some surprising predictions for the future. I am grateful for the tremendous effort these experts put into this challenge and hope they stimulate your imagination so you will work to bring these advances to our patients.

Forty Years of Epilepsy Surgery: Where Are We Going?

Dennis Spencer, MD

The evolution of surgery for medically intractable epilepsy began when 19^th century clinicians correlated autopsy pathology with specific seizure disorders (1). Prior to the availability of MRI, the pathological substrate identification was dependent on electroencephalography. Initially, interictal EEG was recorded via intraoperative electrocorticography; later, invasive electrodes were used for ictal recordings, often combining depth, strip, and grid electrodes (2). Once MRI had become an established diagnostic tool, some patients could move directly to surgery without intracranial recordings. However, continuous intracranial recording in the epilepsy monitoring unit (EMU) was still required when the MRI was negative or when it demonstrated injury or malformations of cortical development (MCD). The current surgical decision tree – with the placement of electrodes dependent on noninvasive preoperative data – is illustrated in Figure 1. For most institutions this has been the preoperative surgical paradigm for many years.

Figure 1.

Current surgical decision tree.

Despite advances in technology and surgical techniques, control of seizures in the best of epilepsy surgery centers ranges from 15 to 75 percent and has not changed since the beginning of the MRI era, approximately 30 years ago (3). Clinical and basic research evidence gathered during this same era suggests that aside from tumors and cavernomas, the mechanism of epileptogenesis may involve an epileptic network rather than a single focus.

In support of distributed epileptogenesis in small networks, using depth electrodes, we found independent ictal onset with the same semiology from the hippocampus, entorhinal cortex, and amygdala in the same patient with medial temporal lobe epilepsy (MTLE) (4). At the same time, we know that laser ablating one influential node in the hippocampus may control seizures from a larger network in 50 to 60 percent of patients (5). In the case of bitemporal epilepsy, resecting the most dysfunctional medial temporal lobe provides 50 to 60 percent control, and the same is seen following resection of a normal-volume hippocampus identified with intracranial electrodes as the epileptogenic source (6).

We know there is another unidentified epileptogenic source in approximately 50% of all patients coming to surgery. Multiple papers have confirmed that complete seizure control declines over a 10- to 15-year postoperative period, resulting in 50 to 15 percent of patients remaining seizure free (7).

Dual pathology, defined as mesial temporal sclerosis (MTS) plus either an MCD or developmental tumor, is usually not controlled unless both sources are addressed (8). Finally, neuropsychological testing invariably demonstrates distributed cognitive deficits in most patients other than those with tumors and cavernomas.

Thus, there is ample evidence for distributed epileptogenesis, but the question remains: are those nodes connected and behaving as a network? While abnormalities of connectivity are increasingly described in epilepsy – and in temporal lobe epilepsy in particular – a recent meta-analysis of the literature shows no cohesive results (9).

Evidence from our human-based research has generated some support for epileptogenic networks using three disparate techniques: microdialysis, 7T MR spectroscopy, and electrophysiological connectivity. Extracellular fluid glutamate levels, sampled using microdialysis devices inserted in our depth electrodes in 90 patients, demonstrate elevated basal glutamate levels in the epileptogenic regions (medial temporal or neocortical), but glutamate levels may be even higher in the first propagated region or node after ictal onset (10). Regarding subcortical/cortical networks NAA/Cr levels, an indicator of mitochondrial function and thus energetics can be measured in the human brain using 7T MR spectroscopy. When studied in MTLE with MTS, the epileptogenic hippocampus is energetically depleted, anterior more than posterior. A similar energetic loss is seen in the ipsilateral anterior thalamus and less significantly in the contralateral thalamus and hippocampus (11). Emphasizing thalamic cortical connections, Paz et al. (12) demonstrated that optogenic stimulation of thalamic circuits can inhibit the parietal epileptogenic cortex rendered excitable via a photothrombotic stroke.

We are just beginning to study these potential epileptogenic networks. Modulating them is also very rudimentary, with two open loop and only one closed-loop solution available. Open-loop devices are the vagal nerve stimulator (VNS) and the not-yet-FDA-approved constant stimulation of the anterior thalamic nucleus. The only closed-loop modulatory system is the responsive neurostimulator, which targets dual epileptogenic sites or those in indispensable primary brain regions. It is limited by the number of contacts that can be stimulated and by the fact that it responds to the ictal event only after it has started (13). Clinical trials have demonstrated that each of these devices has about the same efficacy, reducing seizure frequency by less than 50% in 50% of patients. Seizure frequency appears to decline more over time.

Our inability to stop seizures in approximately half of our patients is amplified by our limited ability to normalize the neuropsychiatric comorbidities of depression, anxiety, addictive behavior, and cognitive deficits shared by so many of them.

Researchers are exploring neuromodulation of these same psychiatric diseases in nonepileptic patients by deconstructing the major disorders into functional domains (e.g., fear, reward motivation) and then targeting functional MRI markers of these domains for intracranial stimulation. This research leads us to an intriguing possibility: do any of these neuropsychiatric nodes share the same circuitry as our epileptogenic nodes, and can both networks be modulated together?

To address these questions and to understand epileptogenic networks, we need bioelectric integrated telemetered intracranial monitoring. Our labs are now fabricating a multimodal neuroprobe that records oxygen, intracranial pressure, cerebral blood flow, temperature, and EEG. The next advancement will be the addition of molecular biosensors, which will allow the wireless transmission of critical data. Lastly, we are in the early stages of developing a cooling device that will, temporarily, functionally disable network nodal regions, enabling us to assess cognitive function and the impact on seizure propagation.

In conclusion, the failure of medicine and surgery to evolve toward better seizure control may be the result of the distributed nature of many epilepsy syndromes. Because several lines of evidence connect this distributed pathology into functional and/or structural networks, technological advances make it likely that we can develop bioelectric integrated telemetry studies that will allow a better definition of the networks and a more targeted therapy involving cortical and subcortical structures. Modulation of epilepsy comorbidities is another goal. The new surgical tree will resemble Figure 2.

Figure 2.

Surgical tree incorporating “epilepsy network disorders.”

Harnessing the Power of Bioinformatics in Epilepsy

Tracy Glauser, MD

Bioinformatics is a young science focused on the collection, organization, analysis, and dissemination of data. Its impact spans the entire spectrum of research, from basic to translational to population science. In addition to supporting and enhancing research through collaboration, bioinformatics conducts its own original research with teams of computational and information scientists, biomedical informaticians, and information technologists.

Biomedical informatics has led clinicians and clinical researchers on the journey from raw data to wisdom. The first step was the systematic organization of raw data into accessible information. This started with spreadsheets and then developed into relational databases, allowing scientists to understand previously hidden relationships among the data. The ability to understand patterns within the organized information creates knowledge. Knowledge bases represent the emerging frontier of bioinformatics’ impact on care and research. Ultimately, by understanding the principles within the knowledge gained, one will achieve wisdom. The Data to Information to Knowledge to Wisdom (D → I → K → W) framework is critical to understand the impact and future of bioinformatics in epilepsy.

The rise of bioinformatics technology alone was not enough to accelerate improvements in care and enhance research. A key catalyst was the 2007 Institute of Medicine workshop report focused on “The Learning Healthcare System” (14). The report emphasized that healthcare providers would need help to provide complete evidence-based medicine. It stated that “… although professional judgment will always be vital to shaping care, the amount of information required for any given decision is moving beyond unassisted human capacity” (14). The report issued a call for innovations that would integrate providers with information technology, including clinical decision support systems, universal electronic health records, and tools for database linkage, mining, and use.

This influential report facilitated the development of transformational innovations, including the organization of data from electronic health records (EHR) into usable information. The benefits and challenges brought by the massive implementation of EHR systems are hotly debated, but EHR clearly were the first big step on the clinical D → I → K → W journey.

Building upon these advances, bioinformaticians developed tools like Informatics for Integrating Biology and the Bedside (i2b2) to facilitate discovery on an enterprise-wide scale. These programs allowed reliable and rapid cohort identification, a longstanding obstacle to large-scale clinical research.

The National Patient-Centered Clinical Research Network (PCORnet) is a current innovation that builds upon EHR and data-mining tools. The mission of PCORnet is to “enable faster, more trustworthy clinical research that helps people make informed health decisions” (15). A key part of the underlying strategy was to create a standardized informatics infrastructure, including a common data model. Data sources included EHR data and patient-reported data. This national effort includes 20 Patient-Powered Research Networks (PPRNs) and 13 Clinical Data Research Networks (CDRNs). One of the PPRNs is the Rare Epilepsy Network (REN, https://ren.rti.org/), a collaboration among the Epilepsy Foundation, Columbia University, RTI International, and several patient organizations. PCORnet's 64 DataMarts included data from over 110 million people as of April 2016. Of course, the integration of bioinformatics and patient involvement is not limited to PCORnet. Important current efforts include the development of wearable technology (including home diagnostics), digital therapeutics (gamification wellness), Health Learning Systems, and augmented reality/virtual reality programs.

A collaboration between pediatric epileptologists and computer scientists started in 2003 at Cincinnati Children's Hospital Medical Center with the goal of developing and then implementing epilepsy-focused biomedical informatics innovations to help transform information into knowledge. The first step was to create a machine learning infrastructure. The process started with the selection of the most important clinical measures of epilepsy using a Delphi analysis involving multiple national experts (16). Next was the construction of an ontology that integrated the International League Against Epilepsy's 1989 and 2010 terminologies and concepts with the National Institutes of Neurological Disorders and Strokes’ Common Data Elements. Finally, the team built, validated, and implemented automated methods for feature optimization, annotation, structured and unstructured data extraction and integration, similarity measurement, and physician notification.

This machine learning infrastructure has been implemented to improve epilepsy clinical care in multiple domains. A clinical decision support system for epilepsy (the CHRISTINE system) automatically extracts patient-specific data relevant to medication selection, integrates it using patented algorithms, and produces a hierarchal list of appropriate antiepileptic medications for that specific patient (17). Using a similar approach, the system can extract structured and unstructured patient-specific data and integrate it using natural language processing approaches to determine whether the patient could be a candidate for a specific clinical trial. A third application of the system applies machine learning and natural language processing methods to physician notes to identify potential candidates for epilepsy surgery evaluation. This biomedical informatics approach was able to correctly identify epilepsy surgery candidates earlier than physicians did (18). Lastly, natural language processing methods can be used to identify common epilepsy comorbidities through the analysis of “thought markers,” which capture a patient's neuropsychiatric state through their speech, word choice, and body language. These specific linguistic and acoustic analyses have been able to detect suicidality in adults with a high degree of accuracy (19).

In summary, current bioinformatics innovations can enhance epilepsy clinical care by improving the availability of data and information. Emerging innovations that use natural language processing and machine learning can transform epilepsy clinical care by increasing knowledge and eventually improving wisdom. In the not-too-distant future, daily patient-reported data will be combined with linguistic and acoustic analysis of patient–healthcare provider calls and face-to-face discussions to provide clinicians with a wealth of real-time actionable data. Epilepsy waiting rooms and examination rooms of the future will mirror the audiovisual capabilities of our current epilepsy monitoring units, providing clinicians with real-time data analysis of the patient's thought markers. Harnessing the power of bioinformatics will transform epilepsy care and accelerate research discoveries to help maximize our patients’ potential and quality of life.

Brain Imaging in Epilepsy, Now and in the Future

Jerzy Szaflarski, MD, PhD

The field of neuroimaging is in constant flux because of rapid and dramatic progress over the last few decades. Imaging methods considered experimental today may be standard for epilepsy evaluation in the future. Whether the technology involves structural or functional connectivity, hybrid methods using combinations of MRI with SPECT, PET, or MEG, portable imaging systems, machine learning algorithms, or imaging-based nanoparticle delivery systems to the ictal onset zone, these techniques hold great promise for patients with epilepsy.

In new-onset epilepsies, one study visualized a potentially epileptogenic lesion in 23% of patients while another 22% had nonepileptogenic abnormalities (20). In line with the new-onset seizure guideline, the risk of seizure recurrence in a patient with an epileptogenic lesion is 26% (10% in MRI-negative patients) at 1 year and 48% at 5 years, indicating an at least 2-fold increase in risk of having subsequent seizures in patients with lesions (21). Thus, the presence of an abnormality may not only have a negative effect on the outcome, but may also affect the treatment selection (e.g., earlier surgical evaluation instead of continued pharmacotherapy) because patients with certain types of lesions (e.g., medial temporal lobe epilepsy) are substantially less likely to achieve seizure control with antiepileptic drugs (AEDs) (22). Finally, a recent study has shown that functional connectivity measured with fMRI may be able to predict the course of the disease after the first seizure (23). If this is confirmed in larger prospective studies, fMRI connectivity may become an important tool in the evaluation of new-onset seizures.

While neuroimaging has an unquestionably important role in the evaluation of new-onset seizures, the importance of high-quality imaging increases in patients with treatment-resistant epilepsies. Advances in neuroimaging techniques are most visible in three areas – structural MRI, fMRI, and EEG combined with fMRI (EEG/fMRI). These techniques help determine epilepsy surgery feasibility and the risks of negative cognitive outcomes after resection. Recent technical refinements have improved delineation of brain anatomy and function, both of which are important for surgery outcomes. There is an approximately 2- to 3-fold higher chance of seizure freedom after surgery for patients in whom lesions have been identified (24).

The MRI after the first seizure typically focuses on identifying any potentially deleterious pathologies (e.g., brain tumors or strokes). Subsequent imaging, especially in patients previously deemed MRI-negative, focuses on detecting minute abnormalities, such as bottom of the sulcus dysplasia or minor anatomic changes in the hippocampus (25, 26). Critical factors are the technique used, the resolution of the images, and advanced data processing techniques. Hardware and software advances now allow not only better resolution of the imaged structures (e.g., smaller than 1 × 1 × 1-mm voxel size) to decrease the effects of volume averaging that is especially important for small structures, but also higher-field imaging at, for example, 7T to improve signal-to-noise ratio (SNR) (27, 28). Newer data processing techniques like machine learning algorithms will be used on a daily basis to aid visual inspection. A study of MAP (29), a technique based on voxel-based morphometry, showed a 43% detection rate of previously undetected lesions and showed that resection of the lesion was associated with better outcomes.

Advances in white matter imaging are also important. Techniques like diffusion kurtosis imaging have documented extensive abnormalities not detected by standard diffusion imaging in patients with temporal lobe and other epilepsies (30). Neurite orientation dispersion and density imaging (NODDI) may identify previously undetected small cortical dysplasias (31). While these techniques offer promise in detecting epileptogenic abnormalities in MRI-negative patients, they are awaiting wider implementation and prospective assessments.

While structural connectivity assesses the direct white matter connections between regions or structures, functional connectivity examines the interactions between regions that are not necessarily connected directly (32). The contributions of connectomics to the evaluation of patients with epilepsy are being explored. One study showed that diffusion tensor imaging (DTI) can be used to avoid resection of the Meyer's loop in patients undergoing anterior temporal lobectomy for seizures (33). Finally, structural and functional temporal connectivity were used to predict epilepsy surgery outcomes. (30, 34)

The task-based fMRI studies in epilepsy have initially focused on a comparison between fMRI and intracarotid amobarbital test (IAP) for language and memory lateralization. Although there is a hint that fMRI may be better at predicting postresection verbal memory deficits than IAP, there is insufficient data for fMRI to completely replace IAP (35, 36). More recently, the focus of fMRI studies has shifted to designing tasks that predict long-term cognitive outcomes. In one of the first studies, a semantic decision/tone decision (SDTD) fMRI task showed better sensitivity and specificity than IAP for predicting verbal memory outcomes (37). Several other studies have used different techniques to predict verbal memory decline, including the pictures/words/faces fMRI task, the SDTD task, covert memorization of presented concrete nouns, and a delayed-recognition task (38–41). Another study showed a correlation between stronger lateralization of verbal fluency to the left middle frontal gyrus (MFG) and change in naming scores after resection (42). In summary, the results of these studies indicate that stronger preresection activation lateralization to the dominant hemisphere predicts more decline in verbal memory after dominant resection. However, while fMRI appears to be relatively established in this setting, the procedure has multiple potential drawbacks that continue to dampen clinicians’ enthusiasm for it (36).

EEG/fMRI, contrasts imaging data gathered in the presence and absence of epileptiform discharges, can identify generators of epileptiform discharges in patients with focal and generalized epilepsies (43, 44) and indicate variable involvement of the network in the epileptic process. In generalized epilepsies, EEG/fMRI has provided data in support of the corticothalamic theory of generalized epilepsies and allowed the distinction between idiopathic generalized epilepsies and frontal lobe epilepsies with rapid bilateral synchrony (45, 46). In focal lesional or nonlesional epilepsies, EEG/fMRI results have allowed physicians to proceed with surgical planning and resection in patients who were not considered surgical candidates. In addition, complete resection of the BOLD signal focus related to epileptiform discharges was associated with better surgical outcome compared with incomplete or no resection of this area (47).

Current and Future Trends in the Development of Antiepileptic Drugs

Henrik Klitgaard, PhD

Drug screening in seizure models has for several decades shown a remarkable ability to identify antiepileptic drugs (AEDs) (48). More than 30 of the approximately 40 AEDs introduced worldwide since 1857 have been discovered through their anticonvulsant activity against maximal electroshock or pentylenetetrazol seizures in rodents. This has resulted in the identification of several first-, second- and third-generation AEDs (48). These discoveries have increased treatment options, improved tolerability and safety profiles, and lowered the risk of drug interactions. However, unmet medical need remains in a significant proportion of drug-resistant patients, and there is an absence of treatments that prevent or alter the course of epilepsy (48). This has triggered a transition of the Anticonvulsant Screening Program at NIH into the Epilepsy Therapy Screening Project, whose purpose is to focus future screening efforts on new models, approaches, and technologies that address the unmet need for new and alternative epilepsy therapies.

An important trend in current AED development focuses on approaches containing new pharmacology that can target therapies for specific subpopulations with drug-resistant epilepsy. Hyperactivation of the mechanistic target of rapamycin (mTOR) pathway in epilepsy has attracted attention due to its impact on neuronal excitability (49), and recent findings from a phase III study showed that mTOR inhibition induced a significant reduction in seizure frequency in patients with treatment-resistant tuberous sclerosis complex. Considerable attention has also recently been devoted to the therapeutic potential of cannabidiol in epilepsy and other neurologic disorders. While the primary antiepileptic mechanism of cannabidiol remains to be elucidated, it is unlikely to involve modulation of the endocannabinoid system (50). Instead, it appears to involve a number of voltage-operated ion channels and ligand-gated receptor systems (50). Recent phase III studies showed that cannabidiol provides a significant reduction in seizure expression in patients with treatment-resistant Dravet and Lennox-Gastaut syndromes.

Synaptic, inhibitory GABA_A receptors constitute a main target for existing AEDs (48). Allopregnanolone-like neurosteroids have been shown to be potent allosteric agonists and, at high concentrations, direct activators of both synaptic phasic and extra-synaptic tonic GABA_A receptor–mediated inhibition (51). Ongoing clinical development programs with allopregnanolone and an analogue support the hypothesis that dual inhibition of both synaptic and extra-synaptic GABA_A receptors may translate into a treatment potential for both super-refractory epilepsy and against treatment-resistant seizures in female children with a PCDH19 mutation.

Conventional AED mechanisms involve modulation of voltage- and ligand-gated channels that are incorporated in membranes of neurons (48). In that context, synaptic vesicle protein 2A appears to constitute a novel presynaptic and intracellular AED target. It has been optimized by a recent AED with respect to selectivity, efficacy, and adverse effect profile against treatment-resistant focal-onset seizures (52). This line of research has also generated a family of compounds that possess a dual mechanism that modulates both pre- and postsynaptic inhibition. A recent phase IIa study tested a compound possessing this dual mechanism in adult patients with highly drug-refractory, focal-onset seizures who had failed at least four previous AEDs and who experienced at least four seizures a week. The study reported positive top-line results.

Additionally, recent sequencing technologies have led to an exponential growth in genetic discoveries in epileptic encephalopathies (53). This has enabled the repurposing of drugs from other therapy areas for the treatment of epilepsies induced by mutations in single, causal genes, thereby introducing the concept of precision medicine into epilepsy therapy (53).

The current interest in new pharmacology and gene discovery targeting subpopulations with drug-resistant epilepsy reveals a major change in AED development from a focus on large trials in major patient segments to smaller trials in subpopulations – a transition from a “blockbuster” to “minibuster” model. This promises, in years to come, to yield a new, fourth generation of AEDs that target specific subpopulations of patients with drug-resistant epilepsy.

There appears to be an increasing number of mechanisms that have been reported to possess promising antiepileptogenic properties in preclinical models (48). One such mechanism involves silencing the expression of specific microRNAs (54) and the inhibition of the TrkB kinase pathway, the latter counteracting both status epilepticus-induced spontaneous, recurrent seizures as well as anxiety (55). This line of research reveals that future antiepileptogenic treatments may go beyond small molecule therapies and may also treat various comorbidities.

While there exists a rich platform of animal models of seizures and epilepsy, none of them has been validated to predict antiepileptogenic effects in patients (56). This failure emphasizes the need to carefully assess new mechanisms with valid study designs in different preclinical models and laboratories and to prepare for clinical translation by simultaneous availability of relevant biomarkers (56). For that reason, it is a concern that lack of clinical validation is causing biomarker discovery in epilepsy to lag behind other neurologic specialties. Successful clinical translation is known to be highly dependent on the availability of objective and robust biomarkers (56). For this reason, an alternative antiepileptogenic treatment approach could be to consider combination therapy, with drugs modulating different mechanisms presumed to be involved in the epileptogenic process and shared across a number of acquired epilepsies.

Taken together, the significant progress in recent years in research on epileptogenic mechanisms holds the promise that the future may usher in clinical initiatives that spark the development of novel treatment modalities that may prevent or alter the course of the disease. Specifically, necessary attention should be devoted to facilitating translational research through optimal study design in relevant preclinical models and novel biomarker discovery.

An improved ability to process big data may enhance future development projects associated with the diagnosing and targeting of specific etiologies, while progress in the understanding of autoimmune diseases may permit identification of future curative therapies for autoimmune forms of epilepsy. Finally, advances in devices and algorithms for seizure prediction may enable continuous noninvasive EEG and physiological signal monitoring (57). These technologies could result in a paradigm shift to acute versus chronic treatment for the large proportion of epilepsy patients who have infrequent seizures but receive chronic AED treatment.

Genes and Signaling Pathways: Future Therapeutic Pathways

Peter Crino, MD, PhD

Progress in understanding epilepsy genetics and cell signaling cascades that contribute to epileptogenesis has proceeded at a dizzying pace over the past decade. With the complete sequencing of the human genome and the advent of next-generation gene sequencing technologies, a comprehensive view of causal variants responsible for diverse epilepsy phenotypes and syndromes is within reach. In addition, with widespread availability of sophisticated technologies and bioinformatics tools to assign functional validation to proteins encoded by mutated genes, investigators can rapidly assess how a single gene mutation might affect brain function. These advances will shape the next 20 years of epilepsy research and are likely to yield as-yet unimagined strides in precision medicine for our patients. (58).

Looking back 25 years, one could not have predicted the incredible advances we have made in epilepsy research. In addition to the arrival of myriad new technologies, the discovery of dozens of new epilepsy genes has altered the landscape for understanding epileptogenesis. Over the next generation, our research energies likely will change the landscape for diagnostics, prognostics, discovery, prevention, and targeted therapeutics in epilepsy. Thus, our current perspectives and mandates for research (i.e., eradication of sudden death in epilepsy [SUDEP] or defining a comprehensive database of all epilepsy genes) will dictate perhaps not what will happen, but rather what we should aim to accomplish.

Based on our current technologies in next-generation gene sequencing, it seems probable that over the next 10 to 20 years we will define a comprehensive database of all germline mutations that cause monogenic and syndromic epilepsies. This “epilome” or “epileptome” will include sex-linked, autosomal, and mitochondrial gene loci, and may provide insights into a wide spectrum of malformations of cortical development (MCD), as well as intellectual disability, autism, and psychiatric disorders that vary comorbidly with epilepsy. Furthermore, active functional validation in the laboratory will diminish the number of “gene variants of unknown significance,” and eventually all meaningful variants will be identified and validated. The “epilome” likely will consist of neurotransmitter receptors and uptake sites, ion channels, and synaptic vesicle proteins, and possibly cell adhesion molecules, intrinsic membrane proteins, protein kinases and phosphatases, transcription factors, cytoskeletal and structural genes, and genes involved in cellular transport and metabolism.

The function of the encoded proteins of many of these genes can be gleaned using bioinformatics platforms (i.e., Gene Ontogeny or “GO” software to yield therapeutic target pathways within known cell signaling cascades). Signaling pathways can be targeted for novel drug design or compound library screening. This “precision medicine” approach has already been quite successful in human trials within the mechanistic target of rapamycin (mTOR) cascade (59). Existing and novel compounds can target not only the protein of interest but up- and down-stream modulators and effectors of the target protein. This approach can even be used for dysfunctional ion channels or neurotransmitter receptors; compounds altering the actual function of the channel or receptor can be trialed as well as compounds that affect expression, membrane trafficking, or removal of the site. As the screening technology advances, it seems likely that for rare epilepsies, small cohorts and even “n of 1” trials may be very informative.

From a clinical perspective, it is highly likely within the next two decades that all patients presenting for clinical evaluation will have the opportunity for whole exome (and potentially whole genome) screening to identify single gene mutations that are causative or at least highly associated with epilepsy. Clinical genome sequencing technologies are evolving rapidly. And with a quick turnaround and diminishing costs, whole exome sequencing should be available to all individuals worldwide, yielding a global variant database that could be mined to study mutations within select ethnic, racial, or risk group demographies. A growing literature suggests that tissue samples of a few cells (or even a single cell) may ultimately be adequate for these studies (60).

A logical outcome of widely available whole exome screening will be the usage of genetic data not only to define causal epilepsy variants but also for identification of potential risk alleles for patients at risk for serious consequences of epilepsy such as SUDEP or serious adverse effects of antiepileptic drugs such as Stevens-Johnson syndrome (SJS), liver toxicity, or teratogenicity. For example, there is solid recent evidence that mutations in SCN8A are associated with SUDEP and could be used as screening alleles for patients at risk of sudden cardiac death (61). The identification of the HLA-B*1502 and HLA-A*3101 as high-risk alleles for SJS following carbamazepine use has changed practice approaches for epilepsy patients of Han Chinese descent.

A major challenge facing epilepsy researchers is understanding the pathogenesis of so-called “complex epilepsies,” which have a strong genetic contribution based on family and twin studies but lack a clearly defined, monogenic inheritance pattern (62). Many of these are sporadic epilepsies, including seizures associated with traumatic brain injury (TBI), stroke, brain tumors, or inflammatory disorders, where both genetic and environmental factors may share a role. One compelling idea is that the degree of injury or inflammation in the brain covaries with a rheostatic (i.e., “sliding scale”) network susceptibility to seizures that is unique to an individual's genetic backdrop. Thus, across the genome there could be gene sequence variants (e.g., polymorphisms) in ion channel, neurotransmitter receptor, or cell signaling genes that confer variable susceptibility to epileptogenic effects of brain injury or metabolic derangement. These genes may or may not have a direct link to epilepsy but could confer low, medium, or high susceptibility, or even reduced susceptibility, to seizures. While these variants will likely be identified in protein coding regions (exons), variants in intronic or promoter regions – so-called “junk DNA” – could also be important for gene regulation, silencing, or expression. Susceptibility variants could provide explanations for the variable presentation and clinical phenotype of complex epilepsies as well as the pleiotropy of known genetic epilepsies. For example, an individual suffering from TBI with a complement of variants that confer high susceptibility to seizures might have a more severe phenotype than someone with a similarly severe TBI but a low-risk genotype. Furthermore, implicit genomic susceptibility due to variant distribution could explain widely differing seizure severity in two individuals with tuberous sclerosis complex but similar genotype and number of tubers.

An important consideration in assessing the effects of genomic susceptibility is the effect on seizure risk of the exposome (i.e., trauma, toxins, medications, infections, autoimmune dysregulation, and the microbiome). Over the next 25 years it is highly likely that the combined effects of these environmental risk factors on epileptogenesis will be formalized in large cohorts.

A significant breakthrough in understanding epileptogenesis is the identification of mutations not passed in the germline but instead occurring in dividing somatic cells such as neural progenitor cells during brain development. Mutations in mTOR pathway genes like PI3KCA, AKT, and MTOR itself have been linked to a variety of malformations of cortical development (MCD), including focal cortical dysplasia, hemimegalencephaly, and megalencephaly (63). A fascinating feature of these somatic mutations is that very few cells in the lesions actually contain the mutation. Thus, while the mutation itself may drive the altered cortical structure, other causes such as releasable factors or cell–cell communication may alter migration and lamination in unaffected cell types. A similar paradigm has been proposed in human cancer, where complex genomic tumor microenvironments and mutations can cause cancer growth and progression. Indeed, over time we may find that MCD may contain more than a single mutation at varying alleleic frequencies.

An intriguing idea is the possible existence of a complex somatic mutational environment in many epilepsy subtypes, even in those associated with “normal” brain structure. For example, while we know that somatic mutations can lead to cortical malformations highly associated with epilepsy, an attractive, novel idea is that even so-called nonlesional epilepsies with normal brain architecture could result from somatic mutations in, for example, ion channel or neurotransmitter receptor genes. These mutations would compromise channel or receptor function in a restricted number of cells, and would therefore have significant effects on network hyperexcitability. Identifying somatic change in nonlesional epilepsies could provide new therapeutic targets for drug design and could also yield new clues for preoperative PET or functional imaging approaches that could dramatically help identify seizure-onset zones in nonlesional patients.

A truly exciting opportunity for research will be the engineering of identified somatic mutations into human cell lines in vitro, including neurons, astrocytes, oligodendrocytes, and microglia to provide a strategy to comprehensively evaluate the effects of these variants in the brain. This could be achieved using gene editing techniques such as CRISPR/Cas9 or by generating neurons from induced pluripotent stem cells derived from patient fibroblasts. These cell lines can be used for electrophysiological analysis, cell signaling assays, and for therapeutic drug development. The ability to test compounds on cells containing the exact mutation causing seizures (rather than approximated by knockdown, knockout, or overexpression) would yield a novel approach to precision therapy in human epilepsy.

References

Jackson

, Colman

. Case of epilepsy with tasting movements and “dreamy state”– very small patch of softening in the left uncinate gyrus. Brain 1898; 21: 580–590.

Spencer

, Schwarz

, Spencer

. The treatment of epilepsy with surgery. Merritt Putnam Quarterly 1988; 5: 3–17.

Bulacio

, Jehi

, Wong

, Gonzalez-Martinez

, Kotagal

, Nair

, Najm

, Bingaman

Long-term seizure outcome after resective surgery in patients evaluated with intracranial electrodes. Epilepsia 2012; 53: 1722–1730.

Spencer

, Spencer

. Entorhinal-hippocampal interactions in medial temporal lobe epilepsy. Epilepsia 1994; 35: 721–727.

Gross

, Willie

, Drane

. The role of stereotactic laser amygdalo-hippocampectomy in mesial temporal lobe epilepsy. Neurosurg Clin N Am 2016; 27: 37–50.

Hirsch

, Spencer

, Williamson

, Mattson

. Temporal lobectomy in patients with bitemporal epilepsy defined by depth EEG. Ann Neurology 1991; 30: 347–356.

McIntosh

, Averill

, Kalnins

, Mitchell

, Fabinyi

, Jackson

, Berkovic

. Long-term seizure outcome and risk factors for recurrence after extratemporal epilepsy surgery. Epilepsia 2012; 53: 970–978.

Vale

, Vivas

, Manwaring

, Schoenberg

, Benbadis

. temporal lobe epilepsy and cavernous malformations: surgical strategies and long-term outcomes. Acta Neurochin (Wien) 2015; 157: 1887–1895.

Van Diessen

, Diederen

, Braun

, Jansen

, Stam

. Functional and structural brain networks in epilepsy: what have we learned? Epilepsia 2013; 54: 1855–1865.

10.

Cavus

, Kasoff

, Cassaday

, Jacob

, Gueorguieva

, Sherwin

, Krystal

, Spencer

, Abi-Saab

. Extracellular metabolites in the cortex and hippocampus of epileptic patients. Ann Neurology 2005; 57: 226–235.

11.

Pan

, Kim

, Cohen-Gadol

, Pan

, Spencer

, Hetherington

. Regional energetic dysfunction in hippocampal epilepsy. Acta Neurology Scand 2005; 111: 218–224.

12.

Paz

, Davidson

, Frechette

, Delord

, Parada

, Peng

, Deisseroth

, Huguenard

. Closed-loop optogentic control of thalamus as a tool for interrupting seizures after cortical injury. Nat Neurosci 2013; 16: 64–70

13.

Dalkilic

. Neurostimulation devices used in treatment of epilepsy. Curr Treat Options Neurol 2017; 19: 7.

14.

Institute of Medicine (IOM). Title of article. In: The Learning Healthcare System: Workshop Summary . ( Olsen

, Aisner

, McGinnis

, eds.) Washington DC: National Academies Press, 2007: 5.

15.

The National Patient-Centered Clinical Research Network. PCOR-net 15-minute talk. Available at: http://pcornet.org/wp-content/uploads/2016/06/Hernandez_PCORnet-Fifteen-Minute-Talk-V8_09may2016.pdf Accessed 00 Month, Year.

16.

Pestian

, Matykiewicz

, Holland-Bouley

, Standridge

, Spencer

, Glauser

Selecting anti-epileptic drugs: a pediatric epileptologist's view, a computer's view. Acta Neurol Scand. 2013; 127: 208–215.

17.

Pestian

, Spencer

, Matykiewicz

, Zhang

, Vinks

, Glauser

Personalizing drug selection using advanced clinical decision support. Biomed Inform Insights 2009; 2: 19–29.

18.

Cohen

, Glass

, Greiner

, Holland-Bouley

, Standridge

, Arya

, Faist

, Morita

, Mangano

, Connolly

, Glauser

, Pestian

Methodological issues in predicting pediatric epilepsy surgery candidates through natural language processing and machine learning. Biomed Inform Insights 2016; 8: 11–18.

19.

Pestian

, Sorter

, Connolly

, Cohen

, McCullumsmith

, Gee

, Morency

, Scherer

, Rohlfs

; for the STM Research Group. A machine learning approach to identifying the thought markers of suicidal subjects: a prospective multicenter trial. Suicide Life Threat Behav 2017; 47: 112–121.

20.

Hakami

, McIntosh

, Todaro

, Lui

, Yerra

, Tan

, French

, Li

, Desmond

, Matkovic

, O'Brien

. MRI-identified pathology in adults with new-onset seizures. Neurology 2013; 81: 920–927.

21.

Krumholz

, Shinnar

, French

, Gronseth

, Wiebe

Evidence-based guideline: management of an unprovoked first seizure in adults: report of the Guideline Development Subcommittee of the American Academy of Neurology and the American Epilepsy Society. Neurology 2015; 85: 1526–1527.

22.

Semah

, Picot

, Adam

, Broglin

, Arzimanoglou

, Bazin

, Cavalcanti

, Baulac

Is the underlying cause of epilepsy a major prognostic factor for recurrence?

Neurology 1998; 51: 1256–1262.

23.

Gupta

, Janssens

, Vlooswijk

, Rouhl

, de Louw

, Aldenkamp

, Ulman

, Besseling

, Hofman

, van Kranen-Mastenbroek

, Hilkman

, Jansen

, Backes

. Towards prognostic biomarkers from BOLD fluctuations to differentiate a first epileptic seizure from new-onset epilepsy. Epilepsia 2017; 58: 476–483.

24.

Tellez-Zenteno

, Hernandez Ronquillo

, Moien-Afshari

, Wiebe

Surgical outcomes in lesional and non-lesional epilepsy: a systematic review and meta-analysis. Epilepsy Res 2010; 89: 310–318.

25.

Harvey

, Mandelstam

, Maixner

, Leventer

, Semmelroch

, MacGregor

, Kalnins

, Perchyonok

, Fitt

, Barton

, Kean

, Fabinyi

, Jackson

. The surgically remediable syndrome of epilepsy associated with bottom-of-sulcus dysplasia. Neurology 2015; 84: 2021–2028.

26.

Ver Hoef

, Williams

, Kennedy

, Szaflarski

, Knowlton

. Predictive value of hippocampal internal architecture asymmetry in temporal lobe epilepsy. Epilepsy Res 2013; 106: 155–163.

27.

De Ciantis

, Barba

, Tassi

, Cosottini

, Tosetti

, Costagli

, Bramerio

, Bartolini

, Biagi

, Cossu

, Pelliccia

, Symms

, Guerrini

7T MRI in focal epilepsy with unrevealing conventional field strength imaging. Epilepsia 2016; 57: 445–454.

28.

Santyr

, Goubran

, Lau

, Kwan

, Salehi

, Lee

, Mirsattari

, Burneo

, Steven

, Parrent

, de Ribaupierre

, Hammond

, Peters

, Khan

. Investigation of hippocampal substructures in focal temporal lobe epilepsy with and without hippocampal sclerosis at 7T. J Magn Reson Imaging 2017; 45: 1359–1370.

29.

Wang

, Jones

, Jaisani

, Najm

, Prayson

, Burgess

, Krishnan

, Ristic

, Wong

, Bingaman

, Gonzalez-Martinez

, Alexopoulos

. Voxel-based morphometric magnetic resonance imaging (MRI) postprocessing in MRI-negative epilepsies. Ann Neurol 2015; 77: 1060–1075

30.

Bonilha

, Lee

, Jensen

, Tabesh

, Spampinato

, Edwards

, Breedlove

, Helpern

. Altered microstructure in temporal lobe epilepsy: a diffusional kurtosis imaging study. Am J Neuroradiol 2015; 36: 719–724.

31.

Winston

, Micallef

, Symms

, Alexander

, Duncan

, Zhang

Advanced diffusion imaging sequences could aid assessing patients with focal cortical dysplasia and epilepsy. Epilepsy Res 2014; 108: 336–339.

32.

Middlebrooks

, Yagmurlu

, Szaflarski

, Rahman

, Bozkurt

A contemporary framework of language processing in the human brain in the context of preoperative and intraoperative language mapping. Neuroradiology 2017; 59: 69–87.

33.

Winston

, Daga

, Stretton

, Modat

, Symms

, McEvoy

, Ourselin

, Duncan

. Optic radiation tractography and vision in anterior temporal lobe resection. Ann Neurol 2012; 71: 334–341.

34.

Keller

, Glenn

, Weber

, Kreilkamp

, Jensen

, Helpern

, Wagner

, Barker

, Richardson

, Bonilha

Preoperative automated fibre quantification predicts postoperative seizure outcome in temporal lobe epilepsy. Brain 2017; 140(pt 1): 68–82.

35.

Janecek

, Swanson

, Sabsevitz

, Hammeke

, Raghavan

, Mueller

, Binder

. Naming outcome prediction in patients with discordant Wada and fMRI language lateralization. Epilepsy Behav 2013; 27: 399–403.

36.

Szaflarski

, Gloss

, Binder

, Gaillard

, Golby

, Holland

, Ojemann

, Spencer

, Swanson

, French

, Theodore

. Practice guideline summary: use of fMRI in the presurgical evaluation of patients with epilepsy: report of the Guideline Development, Dissemination, and Implementation Subcommittee of the American Academy of Neurology. Neurology 2017; 88: 395–402.

37.

Sabsevitz

, Swanson

, Hammeke

, Spanaki

, Possing

, Morris

III , Mueller

, Binder

. Use of preoperative functional neuroimaging to predict language deficits from epilepsy surgery. Neurology 2003; 60: 1788–1792.

38.

Bonelli

, Powell

, Yogarajah

, Samson

, Symms

, Thompson

, Koepp

, Duncan

. Imaging memory in temporal lobe epilepsy: predicting the effects of temporal lobe resection. Brain 2010; 133(pt 4): 1186–1199.

39.

Binder

, Swanson

, Sabsevitz

, Hammeke

, Raghavan

, Mueller

. A comparison of two fMRI methods for predicting verbal memory decline after left temporal lobectomy: language lateralization versus hippocampal activation asymmetry. Epilepsia 2010; 51: 618–626.

40.

Sidhu

, Stretton

, Winston

, Symms

, Thompson

, Koepp

, Duncan

. Memory fMRI predicts verbal memory decline after anterior temporal lobe resection. Neurology 2015; 84: 1512–1519.

41.

Dupont

, Duron

, Samson

, Denos

, Volle

, Delmaire

, Navarro

, Chiras

, Lehéricy

, Samson

, Baulac

Functional MR imaging or Wada test: which is the better predictor of individual postoperative memory outcome?

Radiology 2010; 255: 128–134.

42.

Bonelli

, Thompson

, Yogarajah

, Vollmar

, Powell

, Symms

, McEvoy

, Micallef

, Koepp

, Duncan

. Imaging language networks before and after anterior temporal lobe resection: results of a longitudinal fMRI study. Epilepsia 2012; 53: 639–650.

43.

Kay

, Szaflarski

. EEG/fMRI contributions to our understanding of genetic generalized epilepsies. Epilepsy Behav 2014; 34: 129–135.

44.

Kobayashi

, Bagshaw

, Jansen

, Andermann

, Gotman

, Dubeau

Intrinsic epileptogenicity in polymicrogyric cortex suggested by EEG-fMRI BOLD responses. Neurology 2005; 64: 1263–1266.

45.

, Dubeau

, Gotman

BOLD responses related to focal spikes and widespread bilateral synchronous discharges generated in the frontal lobe. Epilepsia 2015; 56: 366–374.

46.

Szaflarski

, Kay

, Gotman

, Privitera

, Holland

. The relationship between the localization of the generalized spike and wave discharge generators and the response to valproate. Epilepsia 2013; 54: 471–480.

47.

, Fahoum

, Hall

, Olivier

, Gotman

, Dubeau

Electroencephalography/functional magnetic resonance imaging responses help predict surgical outcome in focal epilepsy. Epilepsia 2013; 54: 2184–2194.

48.

Löscher

, Klitgaard

, Twyman

, Schmidt

New avenues for anti-epileptic drug discovery and development. Nat Rev Drug Discov 2013; 12: 757–776.

49.

Meng

, Yu

, Song

, Chi

, Tan

Role of the mTOR signaling pathway in epilepsy. J Neurol Sci 2013; 3332: 4–14.

50.

Bih

, Chen

, Nunn

AvW

, Bazelot

, Dallas

, Whalley

. Molecular targets of cannabidiol in neurological disorders. Neurotheraputics 2015; 12: 699–730.

51.

Reddy

, Estes

. Clinical potential of neurosteroids for CNS disorders. Trends Pharmacol Sci 2016; 37: 543–561.

52.

Klitgaard

, Matagne

, Nicolas

, Lamberty

, De Ryck

, Kaminski

, Leclercq

, Niespodzany

, Wolff

, Wood

, Hannestad

, Kervyn

, Kenda

Brivaracetam: rationale for discovery and preclinical profile of a selective SV2A ligand for epilepsy treatment. Epilepsia 2016; 57: 538–548.

53.

Striano

, Vari

, Mazzocchetti

, Verrotti

, Zara

Management of genetic epilepsies: from empirical treatment to precision medicine. Pharmacol Res 2016; 107: 426–429.

54.

Jimenez-Mateos

, Engel

, Merino-Serrais

, McKiernen

, Tanaka

, Mouri

, Sano

, O'Tuathaigh

, Waddington

, Prenter

, Delanty

, Farrell

, O'Brien

, Conroy

, Stallings

, DeFelipe

, Henshall

. Silencing microRNA-134 produces neuroprotective and prolonged seizure-suppressive effects. Nat Med 2012; 18: 1087–1094.

55.

Liu

, Gu

, He

, Joshi

, Wackerle

, Rodriguiz

, Wetsel

, McNamara

. Transient inhibition of TrkB kinase after status epilepticus prevents development of temporal lobe epilepsy. Neuron 2013; 79: 31–38.

56.

Simonata

, Löscher

, Cole

, Engel

, Kaminski

, Loeb

, Scharfman

, Staley

, Velisek

, Klitgaard

Finding a better drug for epilepsy: preclinical screening strategies and experimental trial design. Epilepsia 2012; 53: 1860–1867.

57.

Kim

, Salvatore

, Araki

, Chiaelli

, Xie

, Banks

, Sheng

, Liu

, Lee

, Jang

, Heo

, Cho

, Luo

, Zimmerman

, Kim

, Yan

, Feng

, Xu

, Fabiani

, Gratton

, Huang

, Paik

, Rogers

. Battery-free, stretchable optoelectronic systems for wireless optical characterization of the skin. Sci Adv 2016; e1600418.

58.

Barker-Haliski

, Friedman

, White

, French

. How clinical development can, and should, inform translational science. Neuron 2014; 84: 582–593.

59.

French

, Lawson

, Yapici

, Ikeda

, Polster

, Nabbout

, Curatolo

, de Vries

, Dlugos

, Berkowitz

, Voi

, Peyrard

, Pelov

, Franz

. Adjunctive everolimus therapy for treatment-resistant focal-onset seizures associated with tuberous sclerosis (EXIST-3): a phase 3, randomised, double-blind, placebo-controlled study. Lancet 2016; 388: 2153–2163.

60.

Gawad

, Koh

, Quake

. Single-cell genome sequencing: current state of the science. Nat Rev Genet 2016; 17: 175–188.

61.

Veeramah

, O'Brien

, Meisler

, Cheng

, Dib-Hajj

, Waxman

, Talwar

, Girirajan

, Eichler

, Restifo

, Erickson

, Hammer

. De novo pathogenic SCN8A mutation identified by whole-genome sequencing of a family quartet affected by infantile epileptic encephalopathy and SUDEP. Am J Hum Genet 2012; 90: 502–510.

62.

Thomas

, Berkovic

. The hidden genetics of epilepsy-a clinically important new paradigm. Nat Rev Neurol 2014; 10: 283–292.

63.

Crino

. The mTOR signalling cascade: paving new roads to cure neurological disease. Nat Rev Neurol 2016; 12: 379–392.