Systems biology of human epilepsy applied to patients with brain tumors


  • Sandeep Mittal,

    1. Department of Neurosurgery, Wayne State University, Detroit, Michigan, U.S.A
    2. Karmanos Cancer Institute, Wayne State University, Detroit, Michigan, U.S.A
    Search for more papers by this author
  • Aashit K. Shah,

    1. Department of Neurology, Wayne State University, Detroit, Michigan, U.S.A
    Search for more papers by this author
  • Daniel T. Barkmeier,

    1. The Center for Molecular Medicine and Genetics, Wayne State University, Detroit, Michigan, U.S.A
    Search for more papers by this author
  • Jeffrey A. Loeb

    Corresponding author
    1. The Center for Molecular Medicine and Genetics, Wayne State University, Detroit, Michigan, U.S.A
    2. Department of Neurology and Rehabilitation, University of Illinois at Chicago, Chicago, Illinois, U.S.A
    • Address correspondence to Jeffrey A. Loeb, Department of Neurology and Rehabilitation, University of Illinois at Chicago, NPI North Bldg., Room 657, M/C 796, 912 S. Wood Street, Chicago, IL 60612, U.S.A. E-mail:

    Search for more papers by this author


Epilepsy is a disease of recurrent seizures that can be associated with a wide variety of acquired and developmental brain lesions. Current medications for patients with epilepsy can suppress seizures; they do not cure or modify the underlying disease process. On the other hand, surgical removal of focal brain regions that produce seizures can be curative. This surgical procedure can be more precise with the placement of intracranial recording electrodes to identify brain regions that generate seizure activity as well as those that are critical for normal brain function. The detail that goes into these surgeries includes extensive neuroimaging, electrophysiology, and clinical data. Combined with precisely localized tissues removed, these data provide an unparalleled opportunity to learn about the interrelationships of many “systems” in the human brain not possible in just about any other human brain disorder. Herein, we describe a systems biology approach developed to study patients who undergo brain surgery for epilepsy and how we have begun to apply these methods to patients whose seizures are associated with brain tumors. A central goal of this clinical and translational research program is to improve our understanding of epilepsy and brain tumors and to improve diagnosis and treatment outcomes of both.

Epilepsy Surgery Generates Large Datasets on Many “Systems”

One of the most difficult decisions to make is determining which portions of the brain to surgically remove to treat focal forms of refractory epilepsy. At the present time, surgery is still the only treatment that can cure epilepsy, since current medications are needed continuously and can only suppress seizures. Given the complexity of the human brain and the fact that each person's seizure disorder is unique, involving different anatomic areas and having distinct electrical patterns, it is critical to define precisely the areas of the brain that contribute to seizure generation and balance this with preservation of normal brain function. Often this requires a two-stage surgical procedure that involves placement of intracranial recording electrodes either directly on the brain surface or deep within the brain. As a result, more types of data are generated about the human brain for a patient undergoing epilepsy surgery than for just about any other brain disorder.

Prior to surgery, patients have had a presurgical workup that includes a detailed history and physical examination, neuropsychological testing, magnetic resonance imaging (MRI), electroencephalography (EEG), and long-term video-EEG monitoring of seizures. In addition, many patients also have positron emission tomography (PET) studies, single photon emission computed tomography (SPECT) studies, functional MRI (fMRI) testing, and Wada testing. These tests are used to define the regions where seizures start as well as areas of normal brain function; however, when the seizures are close to eloquent areas or difficult to localize from surface EEG recordings, intracranial electrodes are usually placed for 3–5 days or more. Often >100 electrodes are placed in and around presumed seizures focus areas, as shown in Figure 1. Long-term intracranial monitoring with functional mapping procedures are used to determine the spatial relationship between these areas and areas that support normal eloquent brain function. This information is then used to design a second surgical procedure to remove epileptic tissues while maximally preserving brain function.

Figure 1.

Systems biology of human epilepsy. Patients who undergo a two-stage surgery for treatment of refractory epilepsy often require subdural grids to localize epileptic activity as well as to map normal cortical function. These recordings reveal both seizures as well as interictal spikes (labeled green) that can be quantified and plotted back on the patient's brain surface map in three dimensions where regions of highest spiking are colored red. Each region of tissue removed to treat the epilepsy thus can be associated with a specific anatomic location and spiking/seizure pattern, and can be linked to histology and a number of molecular measures including genes (genomics), proteins (proteomics), and other small molecules (metabolomics).

This highly laborious and comprehensive approach not only generates the best surgical outcome, but also generates a diverse dataset of multiple systems. These include anatomic and metabolic data from imaging studies, electrical data from both scalp and brain surface and depth recordings, and clinical and neuropsychological data. Because the goal of this procedure is ultimately to remove epileptic tissues, epilepsy surgery offers unique research opportunities to relate all of these in vivo systems to one another and to those generated from removed tissues precisely linked to the same brain regions (Fig. 1). In addition, a part of the brain that may not have been involved in seizure onset may also be removed during surgery as part of routine clinical care. This brain tissue can provide invaluable internal “controls” for the same patient.

Systems Biology of Human Epilepsy

Systems biology is variously defined, but one definition is “the ability to obtain, integrate and analyze complex data from multiple experimental sources using interdisciplinary tools.” Our group has been taking advantage of the diverse datasets generated in the surgical treatment of patients with epilepsy by integrating clinical datasets with molecular ones derived from resected tissues (Loeb, 2010, 2011). Because of the computational nature of systems biology, it is important that data be recorded quantitatively. In fact, it is surprisingly how little quantitative analysis is performed at most epilepsy surgery centers. For example long-term electrocorticography recordings are not commonly quantified. Toward this end, we have developed a number of algorithms to quantify both seizures and interictal spiking from intracranial recordings as shown in the electrocorticography recording in Figure 1, where the interictal spikes are marked (Yadav et al., 2011, 2012; Barkmeier et al., 2012b). Once these files are marked, and a trained EEG reviewer has confirmed that the algorithm is marking the EEG correctly, the data are then plotted onto the three-dimensional brain surface of the patient's own brain built from high-resolution MRI images (Beaumont et al., 2012). This generates quantitative data that are linked to brain anatomy. Not only is this useful for research purposes, but this procedure also sets a higher standard for clinical data that are critical for decision-making about what to remove surgically. Application of these methods to patients with brain tumors and epilepsy is described below.

Using Systems Biology to Identify New Pathways, Biomarkers, and Drug Targets

Quantifying multiple data types from epilepsy surgery patients is of critical importance to optimize patient care. It also provides a quantitative background to ask fundamental questions about why some brain regions are epileptic and others are not. Although there are many known lesions associated with epileptic disorders, often the histopathology at the sites where seizures start and/or with high levels of interictal spiking are found to be essentially normal (Seo et al., 2009). Given the revolution in genomics and other high-throughput measures of biologic molecules, it is now possible to measure simultaneously thousands of molecular variables in resected human brain tissues. With our systems biology approach, these can be linked precisely to any number of anatomic, electrical, or clinical parameters through a relational database (Loeb, 2010). An example of how this approach has led to the discovery of genes associated with both seizure onset zones and regions of high interictal spiking in the neocortex is shown in Fig. 2.

Figure 2.

New biomarker and drug target identification from systems biology. Using high-throughput genomic methods such as microarray slides with thousands of specific gene sequences, it is possible to identify genes that are differentially expressed between pairs of tissue samples from within the same patient's brain. On the left is an example where the genes identified are associated with seizure activity and on the right are genes that are changed in response to high levels of interictal spiking. Each of these measures can reveal a highly consistent group of genes common to many patients. Of interest, there is a significant overlap between seizure onset and high spiking genes, but also genes unique to each electrical pattern. Common patterns of gene expression are then used to identify biomarkers for epilepsy as well as new targets for drug development.

Most of the experimental designs we have used ask a simple question: what is different between two or more nearby brain regions within the same patient's brain subjected to long-term intracranial recordings and for which tissue was removed as part of the patient's therapeutic surgical procedure. Invariably, when adults and children with refractory epilepsy undergo large resections, there are regions of the brain that show more or less epileptic activities and can thus be compared to one another. One hypothesis we have tested is that common changes in gene transcription (the levels of messenger RNA [mRNA] molecules for each gene) exist for all patients with epilepsy that share a common electrical pattern. Specifically, we have taken advantage of this internally controlled system to ask what is unique about brain regions where either seizures start or with high levels of interictal spiking (Rakhade et al., 2005, 2007; Rakhade & Loeb, 2008; Beaumont et al., 2012; Lipovich et al., 2102). Both experimental designs shown in Figure 2 generate a group of differentially expressed genes. Not surprisingly, a significant proportion of these genes are shared in both seizure and spiking brain regions. This might be expected, as seizure-onset regions often have high levels of interictal spiking as well (Asano et al., 2003). In addition to mapping the expression of genes that code for proteins as a means to understand the cellular building blocks that produce epileptic circuits, we have also explored a large number of genes that do not code for proteins, but produce long noncoding RNAs that have been shown to have important regulatory functions (Lipovich et al., 2102). Finally, and perhaps even more interesting, we discovered genes that there are unique to seizures and genes unique to spiking, suggesting that these two electrical phenomena have distinct transcriptional mechanisms and hence may require separate treatments (Fig. 2).

Once a common pattern of genes that are turned up and down are identified, statistical or bioinformatic methods are used to identify putative pathways (Beaumont et al., 2012; Lipovich et al., 2102) and cellular differences in the tissue (Dachet F, Bagla S, Keren-Aviram G, Morton A, Balan K, Kupsky W, Song F, Dratz E, Loeb JA, unpublished results). These pathways and cellular differences can then be used to develop an understanding of epilepsy as well as new research directions for a given form of human epileptic activity. In the case of interictal spiking, many of the identified genes vary as a function of spike frequency, suggesting either that the genes are induced by the activity or that the level of gene expression somehow mediates the frequency of spiking (Rakhade et al., 2007). One of the most salient pathways that has emerged from this work is that interictal spiking is closely linked to an intracellular signaling pathway called MAP kinase (MAPK) (Rakhade et al., 2005; Beaumont et al., 2012). In the human neocortex, reagents that “stain” for this pathway were used as tissue “biomarkers” to localize it to the most superficial cortical layers in epileptic brain regions. Of interest, despite a normal-appearing histology, these same superficial brain regions had a significant increase in synaptic density, raising the hypotheses that interictal spiking results from an increase in the synaptic connectivity of superficial cortical layers and that this is mediated through MAPK signaling pathways. In order to test this hypothesis, we developed an animal model of interictal spiking using tetanus toxin in the rat, and not only found the same pattern of biomarker activation in superficial neocortical layers, but also found that when this pathway is blocked using a MAPK inhibitor, we can prevent interictal spiking (Barkmeier et al., 2012a). This approach from human tissue genomics to biomarkers to therapeutic targets reveals the true power of systems biology.

Brain Tumors Have Varying Degrees of Epileptogenicity

A significant proportion (30–50%) of patients with brain tumors have seizures, and many have seizures as their presenting symptoms, especially in the case of low-grade gliomas (van Breemen et al., 2007; Mittal et al., 2008). Some of these seizures become stereotypical and recurrent, thus meeting criteria for the diagnosis of epilepsy. Exactly why brain tumors produce seizures and epilepsy is not known, however, the clinical presentation of the seizure disorder can vary with the type of tumor, the rate of tumor growth, and the location of the tumor in the brain. For example, tumors that involve deeper, subcortical structures are significantly less epileptogenic than those involving the neocortex and hippocampus (Mittal et al., 2010). Quite often the slower the tumor grows, the more likely the patient is to have epilepsy with recurrent seizures, often refractory to standard anticonvulsant medications. Not surprisingly, patients with both slow growing as well as highly aggressive tumors are treated surgically to remove the tumor as much as possible while preserving eloquent brain regions. What is less clear is the best approach to treat both the tumor and the epileptic disorder. For example, although a total lesionectomy produces better seizure control than subtotal lesionectomy does, it is often not completely curative (Chang et al., 2008; Englot et al., 2011a,b). This is of particular importance for children and adults with developmental or slow-growing tumors who will survive their brain tumor surgeries, but may or may not improve their seizure disorders.

Treating Epilepsy and Brain Tumors at the Same Time

As a means to understand the relationship between epilepsy and brain tumors, and to treat both, we have surgically implanted intracranial recording electrodes and obtained long-term recordings as part of a two-stage procedure (Mittal et al., 2010). Figure 3 shows two examples where we have localized regions where seizures start from subdural grid recordings on two patients with brain tumors shown under the cortical surface. Each electrode involved in seizure onset is colored red, whereas seizure-free zones are colored green. As might be expected, seizures in patient A start in the neocortex overlying the tumor. In patient B, however, seizure-onset zones exist both overlying, as well as at a distance from the indicated tumor boundaries. In patient B, if only the tumor is resected and not all of the seizure-onset zones, there is a significant chance that the seizure disorder would not improve after surgery, even though the tumor may be adequately resected.

Figure 3.

Applying systems biology tools to patients with brain tumors. Because many patents with brain tumors have epilepsy as well, we have used our systems biology approach to examine the relationship between brain tumors and epileptic activities. Although the patient shown in (A) has an epileptic seizure focus lying directly over the brain tumor, the patient in (B) has seizure-onset regions both over and at a distance from the tumor margin. A standard resection of the lesion would be expected to also treat the patient's seizures in A; however, a similar lesionectomy for patient B would not remove all the seizure onset regions and hence might be unlikely to cure the seizures.

We are currently studying these relationships not only to identify seizure-onset zones, but also for interictal spiking with respect to the tumor anatomy for a variety of tumor types, tumor locations, and tumor growth rates. It is hoped that a detailed, quantitative analysis of the electrical properties of the brain in and around brain tumors will improve the seizure outcomes in patients with brain tumors. It is also possible that by looking at these electrical patterns, we will improve outcome and survival from the tumor, as the electrical properties of the tissue reveal microscopic extension of the tumor not well visualized on imaging studies.


The authors have no conflicts of interest to disclose. The authors confirm that they have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.