|Home | About | Journals | Submit | Contact Us | Français|
Many symptoms of neurologic or psychiatric illness—such as cognitive impairment, depression, anxiety, attention deficits, and migraine—occur more frequently in people with epilepsy than in the general population. These diverse comorbidities present an underappreciated problem for people with epilepsy and their caregivers because they decrease quality of life, complicate treatment, and increase mortality. In fact, it has been suggested that comorbidities can have a greater effect on quality of life in people with epilepsy than the seizures themselves. There is increasing recognition of the frequency and impact of cognitive and behavioral comorbidities of epilepsy, highlighted in the 2012 Institute of Medicine report on epilepsy. Comorbidities have also been acknowledged, as a National Institutes of Health (NIH) Benchmark area for research in epilepsy. However, relatively little progress has been made in developing new therapies directed specifically at comorbidities. On the other hand, there have been many advances in understanding underlying mechanisms. These advances have made it possible to identify novel targets for therapy and prevention. As part of the International League Against Epilepsy/American Epilepsy Society workshop on preclinical therapy development for epilepsy, our working group considered the current state of understanding related to terminology, models, and strategies for therapy development for the comorbidities of epilepsy. Herein we summarize our findings and suggest ways to accelerate development of new therapies. We also consider important issues to improve research including those related to methodology, nonpharmacologic therapies, biomarkers, and infrastructure.
Over the past decade, there has been increasing recognition that epilepsy is a spectrum disorder that is defined by spontaneous recurrent seizures, but for many people, epilepsy also includes a number of comorbid conditions that substantially affect their health and quality of life. These include cognitive impairment, depression, anxiety, attention deficit, autism, suicidality, psychosis, and migraine. Health concerns related to cardiac function, obesity, and sleep are also common. These diverse comorbidities present a tremendous problem for people with epilepsy and their caregivers, decreasing the quality of life, complicating treatment, and increasing mortality. The 2012 Institute of Medicine (IOM) Report, entitled “Epilepsy Across the Spectrum,” emphasized this point: “Individuals with epilepsy are at risk not only for seizures, but also for a myriad of comorbid health conditions…. Often the comorbidities that accompany epilepsy outweigh the burden of the seizures themselves” (Gilliam et al., 2003; Institutes of Medicine Committee on Public Health Dimensions of the Epilepsies, 2012). Despite the increasing recognition of the frequency and impact of cognitive and neuropsychiatric comorbidities of epilepsy, highlighted in the 2012 IOM report and the National Institutes of Health (NIH) Benchmarks, relatively little progress has been made in developing new therapies directed specifically at comorbidities.
However, there have been considerable advances in understanding the underlying mechanisms that contribute to comorbidities in epilepsy. For example, there is increasing evidence for shared bidirectional pathophysiology between depression and epilepsy (Kanner, 2011; Adelow et al., 2012; Hesdorffer et al., 2012). Studies in animal models have demonstrated shared mechanisms and genetic predispositions between seizure susceptibility and behavioral phenotype (Groticke et al., 2007; Tabb et al., 2007; Groticke et al., 2008; Kim et al., 2009; Mazarati et al., 2010; Pineda et al., 2010; Sarkisova & van Luijtelaar, 2011; Takechi et al., 2011). In addition, there have been advances in the identification of potential targets for therapy and prevention (Johnson & Shorvon, 2011; Mula, 2012). As part of the International League Against Epilepsy/American Society for Epilepsy (ILAE/AES) workshop on comorbidities, our working group addressed the current state of understanding of the comorbidities. Herein we define terminology related to studies of comorbidities, methods, and we suggest ways to accelerate development of new therapies. We also consider issues specific to preclinical research including those related to study design, evaluation of nonpharmacologic therapies, biomarkers, and infrastructure.
The following terms are commonly used when discussing comorbidities, but are often used with different meaning. Therefore, we suggest definitions for future use that can potentially unify the field and avoid confusion.
A condition that occurs in association with another (i.e., epilepsy) at frequencies that are significantly greater than those observed in the appropriate control group. A comorbidity may be a cause of epilepsy, a consequence of epilepsy, or a separate condition that is associated with epilepsy because there is a common cause for the epilepsy and the comorbidity. It is important to note that in the future, some of these comorbidities could be shown to result from a spurious association, in which case they would not be true comorbidities.
A cause is a factor that contributes to the occurrence of a disease or other outcome. Manipulation of the cause leads to a corresponding change in that outcome. The link between cause and effect may be direct or indirect. The understanding of the connection between cause and effect may occur at many levels, from societal to molecular. The appropriateness of each level depends on the purpose of the investigation and the type of explanation sought or intervention planned. Causes may be (1) necessary (i.e., the presence of the “cause” is mandatory for the effect), (2) sufficient (i.e., the cause is all that is required to produce the effect), or (3) contributory (i.e., associated with an increased occurrence of the effect but typically neither necessary nor sufficient). A cause may be necessary but not sufficient, or sufficient but not necessary.
This is the series of events and/or mechanisms between an initiating cause and an effect. The more complex the path from the cause to the final outcome, the less direct and weaker is the association between the initiating event and the ultimate effect.
This occurs when two outcomes are independent but a consequence of the same causal factor.
Confounding can occur when two factors (A and B) are statistically related, because they are both associated with a third variable (C). To be confounding in the association between A and B, C must be a cause of A and be correlated with B but not be part of the causal pathway between A and B. C may be a cause of B or just associated with B.
Based on the definitions described earlier, the following relationships between comorbid conditions can be defined:
This phrase refers to two conditions that share an underlying causal factor. For example, both cerebral palsy and epilepsy may be consequences of an early brain insult. Developmental/genetic encephalopathy (described further below) conceptualizes intellectual disability and seizures as being essential comorbidities that are caused by the same underlying condition.
Developmental/genetic encephalopathy: Many factors that cause intellectual disability and related behavioral consequences are associated with epilepsy. It is critical to distinguish between cognitive effects that are due to the underlying cause versus those due to seizures. The concept of a developmental-genetic encephalopathy is that the cause itself produces comorbidities. Tuberous sclerosis complex is an example in which genetic mutations that cause the disorder affect fundamental neurobiologic phenomena (mammalian target of rapamycin [mTOR] signaling) that can lead to cognitive deficits—independent of seizures (Ehninger et al., 2008; van Eeghen et al., 2012a,b).
The term “secondary comorbidity” is used when one condition is involved directly in producing a second condition. The second condition is the secondary comorbidity. For example, epileptic encephalopathy (defined below) conceptualizes intellectual disability and cognitive impairments as secondary comorbidities of seizures. The implication is that effective intervention for one (seizures) may lessen frequency or severity of the other (intellectual disability).
Epileptic encephalopathy: This term refers to a process defined in the 2010 Classification and Terminology report of the ILAE when “the epileptic activity itself may contribute to severe cognitive and behavioral impairments above and beyond what might be expected from the underlying pathology alone (e.g., cortical malformation), and…can worsen over time.” (Berg et al., 2010).
This term is used when the treatment of one condition leads to or exacerbates another condition. For example, many drugs that are used to treat epilepsy can induce cognitive and behavioral impairments.
These terms refer to social–environmental factors that, as a consequence of one condition, may have an impact on an individual and influence the occurrence of another condition. For example, uncontrolled seizures may lead to the loss of driving privileges, which may in turn result in unemployment, social isolation, and depression.
These comorbidities involve multiple mechanisms that may be interdependent. For example, cognitive impairment that is a comorbidity of epilepsy may be produced by seizures or by medications to treat seizures. Furthermore, comorbidities may have different relationships to one another. In some instances, they may be mutually reinforcing. For example, sleep disorders are a common comorbidity of epilepsy. Sleep disorders are comorbid with depression; they may cause depression or be exacerbated by depression.
Animal models provide an opportunity to investigate the pathophysiology of comorbidities of epilepsy, and to explore the temporal relationship between the comorbidity and epilepsy. The temporal relationship is important because some comorbidities begin prior to the onset of epilepsy, whereas others arise after epilepsy is established. Furthermore, animal models provide an opportunity to address the effects of therapeutic interventions (Blumenfeld et al., 2008; Russo et al., 2011).
Consistent with the clinical evidence that many types of epilepsy are associated with an increased incidence of comorbidities, cognitive and behavioral impairments have been reported for diverse animal models of epilepsy. Commonly used rat models of acquired epilepsy, which use kindling, status epilepticus, or traumatic brain injury to induce epilepsy, are accompanied by behavioral abnormalities, including increased anxiety- and depression-like phenotypes (Kalynchuk, 2000; Milman et al., 2005; Mazarati et al., 2007; Jones et al., 2008a) and cognitive deficits (Hamm et al., 1993; Hannesson & Corcoran, 2000; Detour et al., 2005; Kemppainen et al., 2006; Jessberger et al., 2007; Chauviere et al., 2009). The two most commonly studied rat models of genetic generalized epilepsy (GGE) with absence seizures, Genetic Absence Epilepsy Rats from Strasbourg (GAERS) and WAG/Rij rats, also display cognitive impairments and behavioral abnormalities, and these phenotypes have some similarity to psychiatric symptoms of people with GGE (Ott et al., 2003; Sarkisova et al., 2003; Jones et al., 2008a; Sarkisova & van Luijtelaar, 2011), as well as psychosis (Jones et al., 2010). Intriguingly, the behavioral phenotypes of GAERS are manifested before the onset of the seizures, suggesting that the behavioral deficits are not a secondary effect of the epilepsy (Jones et al., 2008a). Many of the behavioral comorbidities in common rodent models of epilepsy are summarized in Table 1.
In addition, there are animal models (described below) that are potentially useful to study comorbidities in epilepsy but are not common in epilepsy research. These “lesser known” models are being used to study psychiatric illness. Some of the seizures in these animals are spontaneous and repetitive, criteria that define epilepsy. Some of the methods to induce epilepsy, such as early life injury, are relevant to human epilepsy (e.g., temporal lobe epilepsy; TLE). Therefore, there is a great opportunity to use these models to study comorbidities in epilepsy.
The definition of comorbidity often depends on the perspective of the investigator: The investigator studying schizophrenia may consider the seizures in the neonatal ventral hypothalamic lesion (NVHL) model (Lillrank et al., 1995; Lipska et al., 1995; Beerpoot et al., 1996), a comorbidity, whereas the investigator interested in epilepsy may consider this animal model to be an excellent animal model of epilepsy where schizophrenia-like symptoms emerge as a comorbidity. More research is needed to clarify which symptoms arise first, the seizures or the psychiatric symptoms, to help address the problem. Until now, the video-electroencephalography (EEG) studies needed for these assessments have not been feasible. Conducting the video-EEG studies in the NVHL model is an example of a way that an NIH benchmark—to create/validate more models of epilepsy with comorbidities—could be addressed. The NVHL model is interesting for other reasons also: One example is that experiments have recently shown that psychiatric symptoms may be modifiable by behavioral interventions during puberty (Lee et al., 2012). Epilepsy researchers can use such studies to provide starting points for potential behavioral interventions that might reduce comorbidities in other animal models of epilepsy, and eventually in humans.
Many investigators are currently interested in shared causes of epilepsy and comorbidities. Supporting the case for a shared causation between the epilepsy and psychiatric comorbidities, rat strains that have an increased predisposition to developing epilepsy following a brain insult (but are not epileptic initially), such as the FAST rats (a strain where kindling occurs rapidly) and genetically epilepsy prone rats (GEPRs), show a range of behavioral phenotypes relevant to the psychiatric comorbidities of patients with epilepsy (Anisman & McIntyre, 2002; McIntyre et al., 2004; Jobe & Weber, 2005; Gilby et al., Runke et al., 2011). In mouse models of Alzheimer’s disease (AD), which often exhibit recurrent seizures and therefore have epilepsy (Palop et al., 2007; Minkeviciene et al., 2009; Scharfman, 2012b), cognitive impairments that accompany seizures may be due to similar mechanisms as the seizures. For example, it has been suggested that one of the mechanisms underlying the cognitive impairments and the seizures are dysfunctional sodium channels, specifically NaV1.1 (Verret et al., 2012). Notably, mutations in NaV1.1 are considered to be the cause of two genetic epilepsy syndromes, genetic epilepsy with febrile seizures plus (GEFS+), and severe myoclonic epilepsy of infancy (SMEI), which is also known as Dravet syndrome, and is accompanied by cognitive deficits (Catterall et al., 2010; Scharfman, 2012a).
The design of any preclinical study that addresses treatment of comorbidities should be clearly outlined. This includes description of the comorbidity (i.e., specific symptom vs. a defined syndrome or disease), the type of comorbidity (essential, secondary, iatrogenic, situational/contextual, or complex/interacting), and the type of the study (e.g., proof of concept, safety/tolerability, biomarker validation).
Several recent publications have outlined recommendations for the design of such studies and guidelines for reporting, to ensure rigor and transparency (Kilkenny et al., 2010; Rigor in Science Workgroup N, 2011; Galanopoulou et al., 2012; Landis et al., 2012). These include: justification of species, strain and experimental model (s), description of testing procedures, justification of drug chosen for study and its formulation, description of drug delivery, description of power analysis and selection of appropriate sample sizes and statistics, randomization procedures, description of inclusion and exclusion criteria, and reporting of missing data. Blinding, although desirable, may be difficult for studies that aim to treat behavioral comorbidities and needs to be stated as a limitation. Unbiased reporting and discussion of all results (negative or positive), discussion of studies supporting or disputing the findings, and disclosure of conflicts of interests are all important. Establishing clinical relevance of preclinical studies, and identifying appropriate target patient populations for studies of epilepsy comorbidity treatment entail some intricacies in design and interpretation that merit special attention, and are discussed below.
The Diagnostic and Statistical Manual of Mental Disorders, 4th edition, Text Revision (DSM-IV-TR; American Psychiatric Association, 2000), provides diagnostic criteria for the majority of psychiatric disorders, based on a consensus opinion reached by a group of experts. However, as our understanding of the clinical semiology of depressive disorders (and other psychiatric comorbidities) has advanced, it has become clear that the proposed classification is incomplete (Kanner et al., 2010). Indeed, in the last decade, psychiatrists have recognized the importance of subtypes of depression (Kanner, 2011).
Most preclinical anticomorbidity studies use behavioral tests that target specific behaviors at predefined time points that may not necessarily parallel clinical diagnostic criteria. These limitations could affect the ability of the preclinical studies to address the clinical disorder. Table 2 exemplifies the complexity of reaching a diagnosis of “depression” in an animal model based on the clinical diagnostic criteria. To adapt the DSM-IV-TR criteria for major depressive disorder for rodents would require longitudinal observations and developing a method—likely to be difficult—to identify the “personality” of a rodent and distinguish normal from depressed personality. In many preclinical studies of epilepsy-related comorbidities, assessment of depressive disorder is based on tests and conditions that vary from laboratory to laboratory. A list of commonly used rodent behavioral tests and what they are used to assess are provided in Table 3. The tests are usually performed at consecutive time points that may span several weeks. In epileptic animals, where the epileptic condition may be evolving, disease progression over several weeks may make the results difficult to interpret. It is also important to note that some tests, like the saccharin preference test or the Porsolt forced swim test (Epps et al., 2012) assess only criteria A1, A2, and perhaps partially A3 of the adapted DSM criteria (Table 2). Furthermore, many of the tasks used to assess affective or cognitive functioning, such as the Porsolt forced swim test, can only be employed once because the task involves stress, which may alter behavior and lead to a different response if the task is conducted a second time. Although such tests have been successful in identifying drugs that improve depressive symptoms in normal animals, examples of pharmacologic studies in epileptic animals are rare. Additional tests such as novelty-suppressed feeding have also been useful to detect effects of selective serotonin reuptake inhibitors (SSRIs) in normal animals (David et al., 2007) but may be difficult in epileptic animals because of the potential effects of caloric restriction on seizures (Hartman et al., 2010). However, if the goal is to identify disease-modifying therapies for comorbidities, it may be important to utilize diagnostic criteria that consider the comorbidity as a disease complex or syndrome rather than defining it by individual symptoms.
An interesting and potentially informative caveat to treatment studies targeting comorbidities in epileptic animals is that drugs used in humans to treat a comorbidity may affect seizures. For example, drugs used for the treatment of depression, such as SSRIs and tricyclic antidepressants, exert antiseizure effects in several animal models of epilepsy. In a model of generalized epilepsy in which generalized tonic–clonic seizures are induced by audiogenic stimuli (the Genetically Epilepsy Prone Rat [GEPR] Jobe et al., 1973), abnormal serotonergic axon arborization has been identified, and deficient postsynaptic 5-hydroxytryptamine receptor 1A (5HT1A) receptor density is present in the hippocampus. Increases in 5HT levels with the SSRI sertraline resulted in a dose-dependent reduction in seizure frequency, which correlated with the serotonin concentration in the thalamus (Yan et al., 1995). In addition, it has been shown that the 5-HT precursor 5-hydroxytryptamine (5-HTP) has anticonvulsant effects in GEPR rats when combined with a monoaminoxidase inhibitor (MAOI) (Jobe et al., 1999). Similarly, a role of 5HT and antiseizure effects of SSRIs has been suggested in multiple focal epilepsy models in several species (Prendiville & Gale, 1993).
For the accurate measures of the signs and symptoms of a disorder, the current approach to behavioral phenotyping may need to be significantly modified. In addition, as mentioned earlier, current approaches often fail to satisfy the requirement of measuring sustained changes in symptoms. The fact that many tasks provide a snapshot of behavior and cannot be repeated illustrates the need to use multiple assays at varied intervals (which themselves could induce affective symptoms) or to develop novel methods. Currently, new computerized approaches are being developed to automate tracking and cataloging of behaviors in the homecage (Jhuang et al., 2010; Lin et al., 2011). Such approaches are noninvasive and could provide more detailed quantification of routine behavior that may be a more sensitive indicator of behavioral symptoms of affective pathology (e.g., sleep disturbances, altered food intake, lethargy, and impaired social behaviors). These methods are in their “infancy,” and will require validation by comparison to existing methods, but could significantly enhance the ability to detect and track behavioral comorbidities.
The selection of species, strains, and genetic background should be based on the availability of accepted animal models with a reproducible clinically relevant phenotype of the seizure-related comorbidity. Validation that spans species, strains, or models would support the value of a treatment, but it would be associated with significant cost, resources, time, and animal use. Interpretation of the results is also complicated in situations where there is incomplete knowledge about the pathogenic mechanisms of a comorbidity or the target relevance in the animal model that is chosen. Therefore, with the exception of iatrogenic comorbidities, for which testing in at least two species (rodent and nonrodent) may be required to meet regulatory requirements, testing many species for early preclinical studies is desired but not required.
On the other hand, validation of the efficacy of a treatment in at least two preclinical studies of the same seizure-related comorbidity, performed in independent laboratories using acceptable standards of testing, is strongly recommended before transition to clinical trials. Such studies could be from the same or different species, strains, or animal models. Incongruent results across studies may not necessarily disprove the findings, if the observed discrepancies can be justified by differences in target relevance or engagement or other identifiable differences in the study protocol or experimental conditions. Reporting conditions used for housing and breeding, genetic background, and procedures for handling animals should also be described.
Seizures, comorbidities, and response to treatments are all influenced by age and sex. Age and sex also influence the interactions between seizures and comorbidities. As a result, age and sex should be addressed in the experimental design of a study. When immature animals are used, age merits additional considerations because epileptogenesis may require time, and be faster or more severe at some ages than others. In addition, epilepsy is not a static condition but dynamic, even in the adult, so superimposing the changing nature of the epileptic condition on the changing nature of a rapidly developing or aging nervous system poses additional challenges. Age-matched controls are needed. Although one sex may be sufficient for a single study, or in cases where a comorbidity occurs only in one sex, testing in both sexes is desired—both for efficacy and tolerability.
One consideration that is important but rarely addressed directly is the interaction of age and sex of the species. In early neonatal life, when many developmental models of epilepsy are studied, circulating levels of 17β-estradiol are critical to normal central nervous system (CNS) development, and this is true for both males and females (McCarthy, 2010; Konkle & McCarthy, 2011). In males, a neonatal surge in testosterone occurs, and testosterone is aromatized to estradiol. Estradiol influences the development of many brain circuits at this early stage of development (McCarthy, 2010; Konkle & McCarthy, 2011). Many of these areas of the brain are affected by seizures or influence seizures, and are likely to influence social behavioral tests (and possibly other behavioral tests as well).
As discussed previously, the relationship between seizures and a comorbidity may be complex, and regulated by many variables, some of which are hard to define. A comorbidity may share risk factors with epilepsy, may directly or indirectly be triggered or accentuated by seizures, their sequelae or their treatments, or may predispose a study subject to have seizures. An optimal study design would anticipate such interactions and control for them. Comparison to appropriate controls to determine the incidence (if any) and/or natural course of the comorbidity in a nonepileptic animal population matched for strain, treatments, handling, and living conditions would be useful. In many cases this detailed characterization may be beyond the scope of an individual study, but relevant discussion regarding the limitations of data would be helpful. Clarifying how treatments affect these interactions may be important in identifying the appropriate patient populations in future clinical trials.
There are three important issues to consider in relation to treatment formulation and administration: (1) the formulation, (2) handling, and (3) dose–response relationships. Regarding the choice of treatment formulation and delivery method, each has the potential to influence or be influenced by the targeted comorbidity. For instance, the use of oral formulations in subjects with reduced food intake (due to loss of appetite) caused by depression may limit drug exposure and underestimate its efficacy. Another example of the complex effects of drug formulation is the effect of drug formulations on a response to placebo, where the drug formulation may generate a placebo effect selectively in one treatment population. For example, sweet oral formulations may cause a placebo effect that is robust in rats with sucrose preference but not in rats without this preference. Handling the animals just before, during, or immediately after delivery may also generate confounding effects, especially when behavioral or cognitive endpoints are evaluated. Therefore, handling should be practiced in the same manner across treatment groups. Dose–response studies are helpful in defining the safety-efficacy profile of the treatment. Definition of the therapeutic window for treatment administration using clinically relevant time points would be most helpful in translating the findings to clinical trials.
Unbiased, blinded, and quantitative methods to monitor outcomes are strongly encouraged. Parallel monitoring of both seizures and comorbid conditions would be helpful to determine whether benefit (or lack thereof) is due to a change in seizures or comorbidities. Parallel assessment of efficacy as well as safety in early stages of therapy development is advised. The goals are (1) to document drug exposure, (2) to define a therapeutic window, (3) to identify health risks that would limit drug development, and (4) to determine the effects of treatment in individuals with epilepsy, because it may differ drastically from the response of the normal population. In studies of disease modification, assessment at least once after there is a drug effect and after drug cessation (“washout”) is desired. To document persistence of a drug effect, assessment at a minimum of two appropriate time points are advisable, with the duration of each evaluation and time between evaluations of sufficient duration to define the effect and document recurrence.
In addition to the general issues of statistical analysis mentioned earlier, the variability of the targeted phenotype within the tested animal population (across animals, transiency of phenotype) should be accounted for in the power analysis and study design. Age, sex, environment, circadian rhythm, genetic background, methods for handling, testing conditions, and seizure frequency should be analyzed as covariates.
As mentioned previously, reporting of both negative and positive effects is essential. In addition, providing evidence for target engagement in the appropriate part of the CNS is critical in distinguishing drug failure from insufficient drug exposure or inappropriate experimental design. Definition of the targeted comorbidity and the subset of patients with epilepsy that exhibit the comorbidity (“target population”) are important. In addition to reporting the effects of treatment on seizures and comorbid conditions, the therapeutic window, nature of treatment (symptomatic vs. disease modifying), and candidate biomarkers would be useful to facilitate translation. Discussion of the barriers to progress to clinical testing would be helpful also. For example, the results of a particular study may suggest a need to consider individualized therapies rather than conventional clinical testing.
In individuals with epilepsy there is an associated high rate of cognitive difficulties that compromise educational progress and achievement during adult life (Davies et al., 2003; Berg et al., 2008). A deficiency in at least one academic area using an IQ achievement–discrepancy definition is identified in about 50% of children with epilepsy (Fastenau et al., 2008), and in between 40% and 60% of children with epilepsy using a low achievement definition (Fastenau et al., 2008). These effects are present even in those functioning at or near an average IQ (Oostrom et al., 2003, 2005). In the school setting, approximately 45% of children with epilepsy and an IQ of at least 80 require special education services, and 16% of these children repeat a year of education (Aldenkamp et al., 1990; Berg et al., 2011). There is increased academic risk in terms of both reading skills (including single word, phonologic awareness, comprehension, and speed or fluency deficits) and math skills (Bailet & Turk, 2000), and often several educational areas are affected in individuals. In addition to these specific academic deficits, children with epilepsy often show general deficits in memory and attention (Metz-Lutz et al., 1999).
The approach to treating a cognitive comorbidity has been primarily through nonpharmacologic interventions such as special education in children and cognitive rehabilitation in adults. The basis for this approach lies in the understanding that there is substantial plasticity in the CNS. Studies over the last decade have demonstrated that neuronal networks are plastic and that connectivity is remodeled throughout life by changes at synapses, as well as other mechanisms (Caroni et al., 2012).
There is evidence in animal models that extensive training (“overtraining”) may ameliorate cognitive impairments in some types of epilepsy. For example, it has been shown that rats exposed to early life seizures exhibit impairments in hippocampal-dependent tasks in adulthood. However, if the rats are overtrained they eventually perform the task as well as controls, which may be related to the ability to increase theta or gamma oscillations (Kleen et al., 2011).
In addition to over-training, environmental enrichment has been effective in reversing spatial learning and memory deficits in rodents following status epilepticus (Rutten et al., 2002). Environmental enrichment prior to status epilepticus reduced apoptosis and the risk of spontaneous seizures after status epilepticus (Young et al., 1999). However, in other animal models, effects of environmental enrichment are varied. In an animal model of atypical absence epilepsy, environmental enrichment improved olfactory recognition and spatial learning, but did not significantly reduce the number of ictal discharges (Stewart et al., 2006). In transgenic mice with a missense mutation of the sodium channel Scn2a (Nav1.2) and recurrent seizures, enriched sensorimotor experience from birth reduced hyperexcitability and histopathologic changes in the mice (Manno et al., 2011). It is important to note that the concept of environmental enrichment, although intriguing, is important to consider carefully. One reason is that most laboratory animals have little enrichment in their typical laboratory environment, which is different from an animal in its natural setting, and different from humans.
The ketogenic diet appears to have a positive effect on cognition in some studies of rodents but not all. For example, in aged rats, the ketogenic diet improved cognition (Xu et al., 2010). In contrast, the ketogenic diet did not appear to influence cognition in normal adult rats (Thio et al., 2010). Zhao et al. (Zhao et al., 2004) found that the ketogenic diet had an adverse effect on spatial learning and memory. However, in amygdala-kindled adult rats, the ketogenic diet did not adversely affect spatial learning or exploratory behavior (Hori et al., 1997).
Depression is a common psychiatric comorbidity in epilepsy. Many hypotheses have been proposed to explain the relationship between depression and epilepsy. For example, Koh et al. (2007) suggested that seizures lead to increased susceptibility to depression by reducing expression of serotonin receptors. This study was an excellent example of an approach to gain insight into a comorbidity using an animal model. The investigators used the kainic acid model, where status epilepticus is induced by systemic administration of kainic acid, and there are chronic spontaneous seizures that develop in the subsequent weeks. The authors induced status epilepticus and then investigated differential gene expression using microarray, reverse transcriptase polymerase chain reaction (RT-PCR) and the Porsolt forced swim test. The authors reported that the gene for one of the serotonin-receptor subtypes was reduced following status epilepticus and that the changes in the gene were paralleled by decreased mobility in the Porsolt forced swim test. Of interest, they also showed that a potential treatment was effective: There was improved behavior in rats reared in an enriched environment. Another study used exploratory behavior to address anxiety. Rats exhibited a reduction in exploration in the open field test after kainic-acid administration. However, rats that were exposed to an enriched environment were similar to controls (Koh et al., 2005). In an animal model of atypical absence epilepsy, environmental enrichment also improved performance on tasks that are used to evaluate anxiety (Stewart et al., 2006).
A biomarker is as an indicator that can be measured objectively and reflects a normal biologic process, the development of disease, or therapeutic intervention (Biomarkers Definitions Working Group, 2001). It should be noted that this definition distinguishes a biomarker from surrogate endpoints that specifically predict clinical outcomes. Within this definition of “biomarker,” genetic, behavioral, anatomic, or metabolic change can all serve as biomarkers, and they can reflect the severity of seizures or a comorbidity, or both.
As biomarkers for comorbidities of epilepsy are developed, it is important that they reflect (1) changes in the disease with time (e.g., progression) and (2) have predictive value for efficacy of therapeutic intervention. It is also valuable if they relate to a mechanism for the comorbidity. Because comorbidities may cause greater impairment in quality of life than seizures do, developing biomarkers of comorbidities of the epilepsies is equally important to developing biomarkers of epileptogenesis. Consideration should be given to biomarkers that reflect both the permanent impairments related to the epilepsy and transient impairments, which may be associated with ongoing seizures and/or medication. Some biomarkers may reflect impairments that contribute to epilepsy (e.g., neuronal damage or loss, synaptic reorganization, inflammation). Other biomarkers, such as interictal spikes and altered behaviors, may be related to dynamic impairments caused by seizures. In considering biomarkers that reflect comorbidities, the following classification is useful. Different types of biomarkers are listed in Table 4, with the acknowledgement that validity of biomarkers (e.g., their predictive value) needs to be defined.
Herein, we define a dyscognitive state as a state that predicts cognitive impairment based on an underlying defect such as a mutation in a specific gene. It is useful when considering “disease biomarkers” versus “symptomatic biomarkers,” where the disease biomarker is a dyscognitive state and the symptomatic biomarker is the symptom (Table 4).
For example, a mutation in MeCP2 gene predicts a dyscognitive state in Rett syndrome. This prediction is useful even if it is employed in the presymptomatic stages. In this case the mutation is a biomarker predicting a dyscognitive state (Chahrour & Zoghbi, 2007; Samaco et al., 2008). In AD, a biomarker detecting amyloid deposits would be a diagnostic tool for determination of dyscognitive state. Examples include in vivo imaging with 11C-PiB-PET amyloid (Driscoll et al., 2012). In patients with mild cognitive impairment (MCI), an increased ratio of ttau/Aβ1–42 (total tau protein/Aβ 1–42) in the cerebrospinal fluid (CSF) may be a biomarker that predicts a dyscognitive state and also predicts progression in AD (Shaw et al., 2011).
Symptomatic biomarkers reflect clinical manifestations of the comorbidity, without necessarily affecting the underlying disease For instance, status epilepticus predicts postictal cognitive impairment. Likewise, electrographic signatures of epileptic encephalopathies associated with cognitive dysfunction are symptomatic biomarkers. Electrical status epilepticus in sleep is a symptomatic biomarker of Landau-Kleffner syndrome (Galanopoulou et al., 2000; Tassinari et al., 2009).
Specific electrographic abnormalities in electrographic status epilepticus or epileptic encephalopathies of infancy may serve as biomarkers that can be used to define important variables. For example, these biomarkers help determine a therapeutic window, because their presence or absence defines efficacy of treatment. They also help guide treatment. Examples include the EEG-guided implementation of barbiturates in status epilepticus (Rossetti & Lowenstein, 2011), hormonal treatments in infantile spasms, or treatment for electrical status epilepticus in sleep (Galanopoulou et al., 2000; Pellock et al., 2010). Another example where biomarkers are used for treatment implementation is the information provided by diagnostic tests to localize the tumor responsible for secreting autoantibodies in autoimmune-mediated paraneoplastic syndromes manifesting in epilepsy (Rosenfeld & Dalmau, 2011).
There are four areas that need to be addressed regarding infrastructure/resources for studies of the comorbidities of the epilepsies. They are (1) validated animal models, (2) training, (3) technology, and (4) funding. First, it is critical to utilize validated animal models for research on the comorbidities of the epilepsies. These models should have characteristics of the human condition they are modeling, and there should be an understanding of the seizure types and epilepsy syndrome that the animals simulate. It is critical that researchers improve the rigor of preclinical studies to maximize the likelihood of success in clinical trials (Landis et al., 2012). In addition, recommendations should be provided for validated protocols so that it is possible to compare results across animal models and laboratories.
Technological advances could greatly enhance research. It would be extremely helpful to develop automated methods to test animals that are less time-intensive, and to develop technology that facilitates long-term EEG monitoring. Developing technology to permit in vivo monitoring of biomarkers would be extremely useful.
There is a significant need for development of interdisciplinary research teams. Epilepsy research teams need to have access to investigators with expertise in testing rodent models and increase training for those who are directly involved in testing the animals.
This research would benefit from an increase in the availability of funds for studies on the comorbidities of the epilepsies, as well as increased interest for support of this type of research by more funding agencies.
Comorbidities in epilepsy can be debilitating, but they are often overlooked. Fortunately, a number of recent efforts are increasing the attention paid to comorbidities. Collaborations between diverse groups with specialized expertise—ranging from cognitive neuroscientists to clinical epileptologists—are making it possible to identify comorbidities earlier, define them more precisely, and develop new treatments. As these new efforts are underway, it will be essential to use terms, methods, and strategies clearly and consistently. Here we have suggested terminology, guidelines for methods, and outline approaches that will help the epilepsy community advance our understanding, diagnostics, and treatment of comorbidities.
The authors would like to thank the American Epilepsy Society (AES), International League Against Epilepsy (ILAE), Citizens United for Research in Epilepsy (CURE), Epilepsy Therapy Project, and Autism Speaks for their financial support of the AES/ILAE Joint workshop in London (September 2012). The authors would also like to thank Patricia Lear and Jennifer Pearson for their invaluable assistance in preparation of this manuscript.
Funding sources: R01NS38595 and grants from CURE, AES, the Epilepsy Foundation (EF) and the Colorado Center for Drug Discovery (ARB-K); NARSAD and EF (KGB); R37NS31146 and a grant from the Pediatric Epilepsy Foundation (ATB); R21NS078333 and R01NS020253 and grants from Autism Speaks, CURE, Heffer and Siegel Families Foundations (ASG); R01NS061991 and CURE (MRW); RO1NS039587, CURE, and Office of the Director grant R21NS072099 (MP); R01NS37562, R21MH090606, the Alzheimer’s Association, and NYS Office of Mental Health (HES).
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.
Dr. Berg has received speaker honoraria and travel support from BIAL and MUSC, and serves on advisory boards for CURE and Eisai. Dr. Galanopoulou has received speaker honorarium from Novartis. Dr. Kanner has served as member of a data safety board for Vertex. Dr. O’Brien has received research grant support from the NHMRC (Australia), the Royal Melbourne Hospital Neuroscience Foundation, Sanofi-Aventis, UCB, SciGen, GlaxoSmithKline, Novartis and Janssen-Cilag, and speaker honorarium from Sanofi-Aventis, UCB, SciGen, GlaxoSmithKline, and Janssen-Cilag. None of the other authors have any conflicts of interest to report related to this manuscript.