research articles on epileptic seizures

Open access
Published: 17 September 2021

The pharmacological treatment of epilepsy: recent advances and future perspectives

Emilio Perucca ORCID: orcid.org/0000-0001-8703-223X 1 , 2

Acta Epileptologica volume 3 , Article number: 22 ( 2021 ) Cite this article

35k Accesses

39 Citations

Metrics details

The pharmacological armamentarium against epilepsy has expanded considerably over the last three decades, and currently includes over 30 different antiseizure medications. Despite this large armamentarium, about one third of people with epilepsy fail to achieve sustained seizure freedom with currently available medications. This sobering fact, however, is mitigated by evidence that clinical outcomes for many people with epilepsy have improved over the years. In particular, physicians now have unprecedented opportunities to tailor treatment choices to the characteristics of the individual, in order to maximize efficacy and tolerability. The present article discusses advances in the drug treatment of epilepsy in the last 5 years, focusing in particular on comparative effectiveness trials of second-generation drugs, the introduction of new pharmaceutical formulations for emergency use, and the results achieved with the newest medications. The article also includes a discussion of potential future developments, including those derived from advances in information technology, the development of novel precision treatments, the introduction of disease modifying agents, and the discovery of biomarkers to facilitate conduction of clinical trials as well as routine clinical management.

The modern treatment of epilepsy started with the introduction of phenobarbital in 1912. The advent of phenytoin in the late thirties marked another milestone, because it was made possible by the introduction of animal models of antiseizure activity [ 1 ]. Similar models also played a key role in the subsequent development of many other antiseizure medications (ASMs). Today, the pharmacological armamentarium against epilepsy includes more than 30 drugs (Table 1 ). These drugs differ in their pharmacokinetics, efficacy, and adverse effect profile, thereby offering unprecedented opportunities to tailor treatment choices to individual needs [ 2 ]. Some of the ASMs introduced after 1985, usually referred to as second-generation drugs, have some safety advantages over older generation agents, but have not increased substantially the proportion of patients who achieve complete freedom from seizures [ 3 ]. For many of these patients, the feasibility of epilepsy surgery, or alternative therapies, should be given early consideration [ 4 , 5 ].

Despite the fact that pharmacoresistance has been little affected by the introduction of newer medications, the drug treatment of epilepsy has made major advances in the last 50 years. In particular, we learnt how to individualize drug selection based on specific patient characteristics such as age, gender, epilepsy syndrome, seizure type, comorbidities, comedications, and other factors affecting clinical response [ 6 ]. We also learnt how to optimize response by careful titration and adjustment of dosage, and use of serum drug levels whenever indicated [ 7 ]. We made major progress in understanding drug interactions, and recognizing the relative merits and indications of monotherapy and polytherapy [ 8 , 9 ]. Likewise, we improved our knowledge of the natural history of epilepsy syndromes, and characterized prognostic factors for seizure recurrence for patients in whom discontinuation of ASMs can be considered after an adequate period of freedom from seizures [ 10 ].

The purpose of the present article is to provide a concise overview of some advances in research on the drug treatment of epilepsy made in the last 5 years, and to discuss currently unmet needs as well as developments which are likely to occur in the foreseeable future.

Advances in characterizing the comparative effectiveness and safety of ASMs

A number of recent clinical trials and observational studies have provided valuable information which can assist physicians in making rational treatment selections. As a follow-up to the initial Standard and New Antiepileptic Drugs (SANAD) trials, which found lamotrigine to be superior to carbamazepine, oxcarbazepine, topiramate and gabapentin in time to treatment failure in patients with mostly focal epilepsy [ 11 ], and valproate to be superior to lamotrigine and topiramate in patients with mostly generalized and unclassified epilepsy [ 12 ], two more recent SANAD trials have been completed. In the first of these trials, 990 adults and children with newly diagnosed focal epilepsy were randomized to receive lamotrigine, levetiracetam or zonisamide, and followed-up for 2 years [ 13 ]. In the per-protocol analysis, lamotrigine was associated with a better 12-month remission from seizures compared with both levetiracetam and zonisamide. In the second SANAD-II trial, which used a similar protocol and enrolled 520 newly diagnosed adults and children, valproate was found to be more effective than levetiracetam in controlling seizures in a pooled cohort of patients with generalized or unclassified epilepsy [ 14 ]. These trials used a pragmatic design mimicking routine clinical practice, even though the possibility of assessment bias due to the open-label, unblinded design cannot be excluded. When considered together with other available lines of evidence, these findings confirm that lamotrigine should be regarded as one of the treatments of first choice for patients with focal seizures. Lamotrigine offers the advantage of being efficacious, generally devoid of adverse effects on mood and cognitive function, and with a low potential to cause adverse drug interactions, even though lamotrigine metabolisn can be affected by a variety of concurrently administered drugs [ 6 ]. One drawback of lamotrigine is the need for gradual titration in order to minimize the risk of serious skin rashes, and therefore it may not be the most appropriate drug for use in patients with frequent severe seizures requiring a prompt onset of antiseizure activity.

The findings from the SANAD studies that valproate is superior to lamotrigine, topiramate and levetiracetam in the treatment of patients with generalized epilepsy are consistent with other lines of evidence. In particular, a recent study from Denmark found that failure to achieve seizure freedom with valproate was the single most important predictor of pharmacoresistance in a cohort of 137 adults with idiopathic (genetic) generalized epilepsy [ 15 ]. The superior efficacy of valproate in controlling seizures associated with generalized epilepsy, however, creates a dilemma in the treatment of females of childbearing potential. In fact, valproate is regarded by the European Medicine Agency as contraindicated for use as first-line treatment in these women (unless the conditions of a rigorous pregnancy prevention programme are fulfilled), due to the higher risk of inducing teratogenic effects as well as impaired postnatal cognitive development in the offspring [ 16 , 17 ].

With respect to teratogenic potential of ASMs, prospective pregnancy registries have contributed greatly to characterize risks associated with individual medications. A particularly important advance was the 2018 publication of data from the international EURAP registry [ 18 ]. This study, based on analysis of 7355 prospective pregnancies exposed to 8 different ASM monotherapies provides risk estimates non only for specific ASMs, but also for different doses of the most commonly used drugs. Overall, the lowest prevalence of major congenital malformations (MCMs) in the offspring was associated with exposure to levetiracetam (2.8% prevalence), lamotrigine (2.9%) and oxcarbazepine (3.0%). Prevalence estimates were intermediate for topiramate (3.9%), carbamazepine (5.5%), phenytoin (6.4%) and phenobarbital (6.5%), and highest for valproate (10.3%). An increased risk with increasing dose was identified for lamotrigine and carbamazepine, and was most prominent for phenobarbital and valproate. In particular, the prevalence of MCMs (with 95% confidence intervals) associated with phenobarbital exposure was 2.7% (0.3–9.5%) at doses < 80 mg/day, 6.2% (3.0–11.1%) at doses > 80 to < 130 mg/day and 11.7% (4.8–22.6%) at doses > 130 mg/day. For valproate, the prevalence of MCMs was 6.3% (4.5–8.6%) at doses < 650 mg/day, 11.3% (9.0–13.9%) at doses > 650 to ≤ 1450 mg/day and 25.2% (17.6–34.2%) at doses > 1450 mg/day. These findings are important because they alert physicians about the need to consider teratogenic risks not only in relation to type of ASM prescribed, but also in relation to dose. A subsequent EURAP investigation documented a clear-cut decrease in the prevalence of MCMs over the period from 2000 to 2013 [ 19 ]. Specifically, there was a 27% decrease in prevalence of MCMs between pregnancies enrolled in the period 2010–2013 compared with those enrolled in the period 2000–2005 (Fig. 1 ). Further analysis of these data provided a strong indication that the improvement in pregnancy outcomes over time was related to changes in ASM prescription patterns, including a major decline in the proportion of pregnancies exposed to valproate. In fact, a reduction in teratogenic risk is one of the important advances associated with the introduction of second-generation ASMs [ 3 ].

Prevalence of major congenital malformations (MCMs) following prenatal exposure to monotherapy with antiseizure medications (ASMs) among cases enrolled in the EURAP international registry during three different periods. Number of exposures during each period (listed in brackets) refer to the eight most common monotherapies, which accounted for 96.7 to 98.1% of all monotherapy exposures. Based on data from Tomson et al [ 19 ].

Introduction of novel ASMs

During the last 5 years, five novel ASMs (brivaracetam, cannabidiol, cenobamate, everolimus and fenfluramine) have been introduced into the market. Key features of each of these medications are summarized in Table 2 . The history of these drugs is illustrative of different strategies being used in developing novel ASMs.

The development of brivaracetam followed a paradigm which has been in place for a very long time, i.e. the structural modification of an already existing medication with the aim of improving its pharmacological profile. Examples of other ASMs developed with this strategy include methylphenobarbital and primidone (both structurally related to phenobarbital), the phenytoin derivative fosphenytoin, and oxcarbazepine and eslicarbazepine acetate, which represent successive modifications of the carbamazepine structure. In fact, levetiracetam itself was originally developed with the aim of improving the pharmacological profile of piracetam, and its antiseizure activity was discovered by chance. Brivaracetam was selected for development after extensive preclinical screening of a large numbers of levetiracetam derivatives. Compared with levetiracetam, brivaracetam has higher affinity for the synaptic vesicle 2A (SV2A), and a similar pharmacological profile [ 20 ]. Brivaracetam has been found to be superior to placebo in adjunctive-therapy trials in focal epilepsy, but its activity profile in other seizure types has not yet been defined in well designed controlled trials. One limitation of brivaracetam is its lack of efficacy when added-on to levetiracetam, presumably due to competition between the two drugs for the SV2A binding site. The still unanswered question about brivaracetam is whether, and to what extent, its efficacy and tolerability profile differs from that of levetiracetam. It has been suggested that brivaracetam is less likely to cause irritability and other behavioral disturbances compared with levetiracetam [ 20 ]. However, evidence on this remains inconclusive, because to date there have been no randomized head-to-head trials comparing these two drugs [ 21 ].

Another ASM recently approved for the treatment of focal seizures is cenobamate, a carbamate derivative which is also structurally related to previously developed drugs [ 22 ]. During clinical development, three confirmed cases of DRESS (Drug Reaction with Eosinophilia and Systemic Symptoms), including one fatality, were reported when cenobamate was titrated rapidly (weekly or faster titration). Consequently, a revised dosing scheme involving initiation at a small dose and slow titration at 2-week intervals has been implemented. In a safety study, no cases of DRESS were reported when 1339 patients were titrated using the slow titration scheme, and the drug has since been approved in the U.S. and Europe [ 23 ]. In the pivotal adjunctive-therapy randomized trial in patients with refractory focal seizures that led to regulatory approval, the most remarkable finding was the relatively high proportion of patients who achieved seizure freedom (21% in the 400 mg/day cenobamate group versus 1% in the placebo group) [ 24 ]. This contrasts with seizure freedom rates ranging from 0 to 6.5% in comparable trials with other second-generation ASMs [ 25 ]. Comparisons of outcome data across trials, however, should be interpreted cautiously, because of differences in clinical settings and characteristics of the patients. Moreover, seizure freedom data in randomized clinical trials refer to limited assessment periods (typically, a 12-week maintenance period) and can be inflated by the last-observation-carried-forward (LOCF) analysis, whereby patients who did not experience seizures but exited the trial prematurely are still counted as seizure-free in the final analysis. In the pivotal cenobamate trial, the proportion of randomized patients who were seizure-free during the entire 12-week maintenance period and did not exit the trial prematurely was 14%, which is still a relatively high proportion [ 24 ].

All three remaining ASMs introduced in the last 5 years share a common feature, i.e. they were approved for orphan indications. Specifically, cannabidiol was approved for the treatment of seizures associated with Dravet syndrome, Lennox-Gastaut syndrome and tuberous sclerosis complex, fenfluramine was approved for the treatment of seizures associated with Dravet syndrome, and everolimus was approved for the add-on treatment of focal seizures in patients with tuberous sclerosis complex (Table 2 ). Introduction of ASMs for orphan indications is a novel development, made possible by increased awareness of the unmet needs associated with many rare epilepsies and by regulatory incentives to develop drugs for these indications, particularly for children. In fact, only 2 of the 25 ASMs developed prior to 2015 were developed for orphan indications, compared with 3 out of 5 ASMs developed in the last 5 years.

Cannabidiol is one of the many active principles contained in the Cannabis plant, which has been used as an herbal remedy in China as early as 2000 BC. Unlike tetrahydrocannabidol (THC), cannabidiol lacks unwarranted psychoactive effects. Its antiseizure efficacy in its approved indications has been demonstrated in several adjunctive-therapy randomized placebo-controlled trials [ 26 ]. In these trials, many patients received concomitant treatment with clobazam, and the improvement in seizure control observed after adding cannabidiol could be ascribed at least in part to a drug interaction. In fact, cannabidiol is a powerful inhibitor of cytochrome CYP2C19 and by this mechanism increases more than three-fold the serum concentration of norclobazam, the active metabolite of clobazam. There is, however, also evidence that cannabidiol retains independent antiseizure activity, unrelated to its interaction with clobazam [ 27 ].

Another ASM recently introduced for the treatment of seizures associated with Dravet syndrome is fenfluramine, which was first marketed in the sixties and widely used in Europe and the U.S. for over 30 years as an appetite suppressant, either as racemic fenfluramine or as its d-enantiomer dexfenfluramine. In 1997, fenfluramine and dexfenfluramine were withdrawn from the market following the discovery of their association with cardiac valvulopathy and pulmonary hypertension. Prior to its withdrawal from the market, however, fenfluramine had been found to improve seizure control in a small cohort of patients with Dravet syndrome, who were allowed to continue treatment with the drug [ 28 ]. These observations led to recent conduction of randomized placebo-controlled adjunctive therapy trials, which demonstrated a robust seizure-suppressing effect in patients with Dravet syndrome, and subsequent regulatory approval for this indication both in Europe and the United States [ 29 ]. Recent data suggest that fenfluramine may also be useful for the treatment of seizures associated with Lennox-Gastaut syndrome [ 30 ]. To date, no evidence of cardiovascular toxicity has been found in patients with epilepsy treated with fenfluramine, possibly because the doses used for seizure protection are generally lower than those used originally for appetite suppression. Yet, there are still many unanswered questions concerning fenfluramine, including the serum levels of parent drug and its active de-ethylated metabolite norfenfluramine required for seizure suppression in children in comparison with those known to be associated with cardiovascular toxicity in adults [ 29 ]. The potential efficacy and safety advantages of developing individual enantiomers of fenfluramine and norfenfluramine should also be considered [ 29 ].

The last medication discussion in this section, everolimus, is an inhibitor of mTOR (mammalian Target Of Rapamycin). Its use in the treatment of focal seizures associated with tuberous sclerosis complex (TSC) has been prompted by evidence linking the pathogenesis of TSC to mTOR overactivation [ 31 ]. Accordingly, everolimus is an example of a novel strategy in drug development, i.e. a precision treatment targeting the etiology of the disease. Everolimus has been found to be effective in reducing tumor growth as well as drug-resistant seizures in TSC patients. Up to 40% of TSC patients show a significant improvement in seizure control when given adjunctive treatment with everolimus. In the pivotal trial that led to its regulatory approval for use as ASM, seizure frequency decreased progressively over time during treatment, suggesting a possible disease modifying effect [ 32 ]. However, a clinically relevant antiepileptogenic or disease-modifying effect has not yet been clearly demonstrated. The age at which treatment is initiated may be important for the final outcome, but controlled trials in children below 2 years of age have not yet been completed [ 31 ].

Cannabidiol, fenfluramine and everolimus are also examples of drugs approved initially for other indications. In fact, cannabidiol was first marketed in a fixed combination product with THC as a nasal spray for the treatment of spasticity associated with multiple sclerosis, fenfluramine was used initially as an appetite suppressant, and everolimus was first approved for the treatment of advanced kidney cancer, subependymal giant cell astrocytoma (SEGA) associated with TSC, pancreatic neuroendocrine tumors, and other tumors. The repurposing for use in epilepsy of drugs initially approved for other indications is one of the options being pursued in the effort to develop precision treatments (see below).

Introduction of novel formulations

Advances in epilepsy treatment can be achieved not only by developing novel drugs, but also by improving the pharmaceutical formulation of already available medications. One area where particularly significant advances have been made in the last 5 years is the development of novel formulations of ASMs for the treatment of seizure clusters and acute repetitive seizures in the out-of-hospital setting [ 33 ]. Until 2018, the only FDA-approved rescue ASM for out-of-hospital use was diazepam rectal gel. In 2019, the FDA approved two additional products for this indication, namely intranasal midazolam [ 34 ] and intranasal diazepam [ 35 ]. Use of these medications is associated with a rapid onset of antiseizure effect, thereby stopping seizures before they progress to established status epilepticus. The intranasal route is generally well accepted by patients and caregivers, as it avoids the social embarrassment associated with use of the rectal route.

Alternative non-rectal rescue formulations of benzodiazepines for out-of-hospital use have been available in other countries for a number of years. In particular, a rapidly absorbed buccal formulation of midazolam has been available in Europe since 2011 [ 36 ]. Innovative formulations of ASMs under development as potential rescue therapy for emergency use include a diazepam buccal film, and an inhaled formulation of alprazolam [ 33 ]. Oral formulations can also be used at times as a rescue treatment, but they can be associated with drawbacks in this setting, such as slower or suboptimal absorption, need for patient cooperation (which is not always feasible) and risk of aspiration [ 36 ].

Future perspectives

Extensive research is ongoing in many areas, and important advances leading to improved epilepsy outcomes are likely to occur in a not too distant future. A few relevant examples will be discussed below.

Increased application of technological tools to epilepsy management

In recent years, information technology (IT)-based applications have been increasingly utilized in epilepsy management, as shown by the widespread use of smartphones to record seizures in the out-hospital setting, and the expanding opportunities offered by Internet-based services in areas such as distant education and telemedicine [ 37 ]. Smartphone applications (apps) are also being increasingly used to assist people with epilepsy to manage and cope with their disease. Most of these apps focus on issues such as treatment management, medication adherence, health care communication, and seizure tracking [ 38 ]. Other apps are aimed at assisting healthcare professionals (HCPs), on example being tools to improve epilepsy diagnosis in non-specialist settings [ 39 , 40 ]. We recently designed a smartphone app to help HCPs in selecting ASM selection for patients with seizure onset at age 10 years or above, particularly in settings where no specialized expertise is available [ 41 , 42 ]. This app is freely available on the Internet ( www.epipick.org ). In a recent validation study, selection of ASMs recommended by the app based on individual patient characteristics was found to be associated with improved seizure outcomes and fewer adverse effects compared with use of ASMs not recommended by the app [ 43 ]. In the future, individualized ASM selection is likely to empowered by more sophisticated technology, including artificial intelligence (machine learning)-based approaches. In a recent study, a machine learning approach combining clinical, genetic and clinical trial data derived from individual patients permitted to construct a computerized model that predicted response to a specific ASM [ 44 ].

In addition to facilitating treatment selection, technology will increasingly assist patients and physicians in monitoring response to treatment through a variety of tools, including seizure detection devices [ 45 ]. Efforts are also ongoing into development of sophisticated technology, including artificial intelligence, for the prediction of epileptic seizures [ 46 ]. This could pave the way to innovative treatment strategies, such as the intermittent use of ASMs prior to the time at which a seizure is predicted to occur.

Precision therapies

In recent years, our understanding of the molecular mechanisms involved in the pathogenesis of epilepsies has improved considerably. One area where advances have been greatest is the genetics of the epilepsies, and in particular the discovery of gene mutations responsible for a large proportion of patients with developmental and epileptic encephalopathies (DEEs) [ 47 ]. The elucidation of an epileptogenic mutation permits to establish the functional abnormality responsible for the epilepsy in the affected individual, and to identify (or develop) precision-therapy medications that may be able to correct such abnormality. One example of a precision therapy is the utilization of the ketogenic diet to control seizures associated with Glucose Transporter Type 1 (GLUT1) deficiency syndrome. In this condition, GLUT1 deficiency results in impaired brain uptake of glucose and consequent neuronal dysfunction, which can be overcome by supplying the brain with an alternative source of energy [ 48 ]. As discussed in recent reviews [ 49 , 50 , 51 ], precision treatments targeting the mechanisms responsible for epilepsy in individuals with specific gene mutations may involve use of drugs previously approved for other indications, a process known as drug repurposing. One example of repurposed drug is the mTOR inhibitor everolimus for the treatment of seizures associated with TSC, as discussed above in this article. Approaches to identify repurposed drugs for specific monogenic epilepsies have been described [ 52 ]. In some cases, improved outcomes can be achieved not by administering additional drugs, but by removing medications that can paradoxically aggravate seizures in these patients [ 49 ]. Importantly, precision therapies are applicable not only to genetic epilepsies, but also to epilepsies due to other etiologies, such metabolic, inflammatory or immune-mediated causes [ 53 ].

At present, application of precision therapies in the management of epilepsy is still in its early days, and will likely expand as further knowledge accrues and newer and more effective targeted treatments are introduced. For genetic epilepsies, targeted (precision) treatments have been reported to improve outcomes in a considerable proportion patients with identified gene mutations [ 54 ], although a more recent survey gave a more sober assessment of the current impact of these treatments [ 55 ].

Biomarker-guided therapies

The search for biomarkers continues to be a hot topic in epilepsy research. Biomarkers can be based on a variety of measures such as genetic, molecular, cellular, imaging, and electrophysiological measures, other clinical or laboratory data, or a combination of these [ 56 , 57 ]. Biomarkers could potentially be used for different purposes, for example to improve diagnostic accuracy, to identify ongoing epileptogenesis and its mechanisms, to predict seizure response (or lack of response) to specific treatments, to assess the probability of seizure recurrence after treatment withdrawal, or to evaluate susceptibility to adverse drug effects. Some biomarkers, such as the HLA-B*15:02 antigen to identify individuals at high risk of carbamazepine-induced serious cutaneous adverse reactions among Han Chinese and other South Asian ethnic groups, are already in routine clinical use [ 58 ].

With respect to potential therapeutic advances, identification and validation of biomarkers could improve treatment outcomes in many ways [ 56 , 57 , 59 ]. First, biomarkers could be used to identify individuals at high risk of developing epilepsy after an epileptogenic insult, thereby permitting selection of these individuals for clinical trials of potential antiepileptogenic therapies. Second, identification of biomarkers predictive of seizure recurrence could facilitate decision on whether to start or withhold ASM therapy in patients who experienced a single seizure. Third, biomarkers predictive of a favorable response to a specific medication would be valuable to select patients to be enrolled in clinical trials of that medication, thereby increasing responder rate and sparing non-responders the burden of receiving placebo or an ineffective treatment. Fourth, biomarkers could theoretically be used to monitor response to treatment, by informing physicians at an early stage on whether the prescribed medication has the required efficacy and safety in a specific individual. Lastly, and most importantly, biomarkers could in the future inform physicians on which ASM is most likely to control seizures effectively and with fewest adverse effects. This may change radically treatment paradigms: for example, the value of a drug which is effective in achieving complete seizure control in 5% of patients with pharmacoresistant epilepsy would be greatly enhanced if we had a biomarker that can identify beforehand those patients who are responsive to that drug. In that scenario, we would use that drug only in responsive patients, thereby increasing the success rate to 100%.

In practice, in most situations it is unlikely that a single biomarker will provide optimal information for any intended purpose. More realistically, breakthroughs are likely to come from algorithms that utilize a combination of biomarkers and other clinical information. The development of artificial intelligence-based tools can facilitate greatly these approaches [ 44 ].

Novel drugs and the search for disease-modifying therapies

The modest impact of second-generation ASMs on seizure outcome in patients resistant to older agents justifies continuing efforts to develop newer and potentially more effective treatments (Table 3 ). Drug development is currently benefiting from many advances, including deeper knowledge of the mechanisms of epileptogenesis and seizure generation in relation to specific etiologies, improved understanding of mechanisms of pharmacoresistance, and availability of disease-specific models as well as models of pharmacoresistance [ 53 , 60 ]. These advances are changing the paradigms used to discover and develop new drugs.

An important paradigm change is a switch from a focus on medications aimed at suppressing seizures to a focus on treatments targeting the underlying disease, i.e. specific etiologies and the molecular mechanisms associated with such etiologies [ 51 , 61 ]. Future precision treatments emerging from this approach will include repurposed drugs [ 50 , 62 ], novel small molecules, and other treatments based on innovative technologies such as antisense oligonucleotides [ 63 , 64 ] and gene therapy [ 65 ]. Some of these therapies require invasive routes of administration, which are also being explored for innovative uses of already established medications [ 66 ].

A closely related paradigm change consists in targeting specifically epileptogenesis and other manifestations of the disease [ 60 , 62 , 67 ]. Such treatment could potentially be used to prevent epilepsy, to inhibit its progression (in those syndromes showing a progressive course), or to alter the appearance or progression of comorbidities such as intellectual disability and other disorders. A wide variety of compounds have been found to possess antiepileptogenic and/or neuroprotective activity in preclinical models through antiinflammatory, antioxidant and other mechanisms [ 68 , 69 , 70 , 71 ]. In addition to novel molecules, these compounds includes naturally occurring substances such as phytocannabinnoids, melatonin, erythropoietin, vitamins and other dietary constituents [ 68 , 70 , 72 , 73 , 74 ], as well as medications already approved for other indications, such as metformin [ 75 ], montelukast [ 76 ], atorvastatin, ceftriaxone, and losartan [ 62 ]. Whether these properties documented in animal models translate into benefit in the clinical setting remains to be demonstrated. Of note, a number of precision treatments directed at specific etiologies of epilepsy could exhibit disease-modifying effects, although it is possible that any medication acting on a single molecular pathway may not address all the complex comorbidities associated with aberrant neural networks [ 61 ].

As discussed above, clinical trials of investigational new drugs could be facilitated by development of biomarkers to detect the occurrence of epileptogenesis at an early stage, to identify drug responsive patients and to monitor response to treatment. Despite claims to the contrary [ 77 ], demonstrating that a chronically administered treatment started before seizure onset prevents the occurrence of seizures in patients still receiving that treatment does not prove epilepsy prevention, because any ASM having a purely symptomatic effect could also produce such an outcome. Likewise, some comorbidities, such as progression of cognitive disability, may be prevented solely as a result of seizure suppression. Truly innovative trial designs will be required to generate unequivocal evidence that a drug is effective in preventing epilepsy, or has a direct disease modifying effect [ 78 ].

Conclusions

Despite the fact that second-generation ASMs have not reduced substantially the burden of pharmacoresistance, advances in the drug treatment of epilepsy continue to be made. These advances result mostly from improved understanding of the comparative efficacy and safety of existing ASMs and from the introduction of newer medicines and innovative formulations. Further advances can be ascribed to technological tools for distant education, telemedicine, and patient empowerment made possible by self-management smartphone-based apps.

It likely that further important therapeutic advances will occur in the coming years. Thanks to ongoing multidisciplinary efforts, clinical outcome for people with epilepsy is likely to improve due to advances in IT technology, development of novel precision therapies, identification of biomarkers to guide drug development as well as routine clinical management, and, ultimately, introduction of truly innovative disease modifying therapies.

Availability of data and materials

Not applicable.

Abbreviations

Antiseizure medication

Developmental and epileptic encephalopathy

Cerebrospinal fluid

Drug Reaction with Eosinophilia and Systemic Symptoms

Dravet syndrome

Cytochrome P450

Electroencephalography

International Registry of Antiepileptic Drugs and Pregnancy

Food and Drug Administration (United States)

Healthcare professional

Gamma-aminobutyric acid

G protein-coupled receptor 55

Information technology

Lennox-Gastaut syndrome

Major congenital malformation

Last-observation-carried-forward

Mammalian Target Of Rapamycin

P-glycoprotein

Standard and New Antiepileptic Drugs

Subependymal giant cell astrocytoma

Synaptic vesicle 2A

Transient receptor potential vanilloid type 1

Tuberous sclerosis complex

Uridine 5′-diphospho-glucuronosyltransferase

Perucca E. Antiepileptic drugs: evolution of our knowledge and changes in drug trials. Epileptic Disord. 2019;21(4):319–29. https://doi.org/10.1684/epd.2019.1083 .

Article PubMed Google Scholar

Moshé SL, Perucca E, Ryvlin P, Tomson T. Epilepsy: new advances. Lancet. 2015;385(9971):884–98. https://doi.org/10.1016/S0140-6736(14)60456-6 .

Perucca E, Brodie MJ, Kwan P, Tomson T. 30 years of second-generation antiseizure medications: impact and future perspectives. Lancet Neurol. 2020;19(6):544–56. https://doi.org/10.1016/S1474-4422(20)30035-1 .

Article CAS PubMed Google Scholar

Dalic L, Cook MJ. Managing drug-resistant epilepsy: challenges and solutions. Neuropsychiatr Dis Treat. 2016;12:2605–16. https://doi.org/10.2147/NDT.S84852 .

Article CAS PubMed PubMed Central Google Scholar

Perucca P, Scheffer IE, Kiley M. The management of epilepsy in children and adults. Med J Aust. 2018;208(5):226–33. https://doi.org/10.5694/mja17.00951 .

Perucca E, Tomson T. The pharmacological treatment of epilepsy in adults. Lancet Neurol. 2011;10(5):446–56. https://doi.org/10.1016/S1474-4422(11)70047-3 .

Patsalos PN, Berry DJ, Bourgeois BF, Cloyd JC, Glauser TA, Johannessen SI, et al. Antiepileptic drugs--best practice guidelines for therapeutic drug monitoring: a position paper by the subcommission on therapeutic drug monitoring, ILAE Commission on Therapeutic Strategies. Epilepsia. 2008;49(7):1239–76. https://doi.org/10.1111/j.1528-1167.2008.01561.x .

Zaccara G, Perucca E. Interactions between antiepileptic drugs, and between antiepileptic drugs and other drugs. Epileptic Disord. 2014;16(4):409–31. https://doi.org/10.1684/epd.2014.0714 .

Schmidt D. Drug treatment strategies for epilepsy revisited: starting early or late? One drug or several drugs? Epileptic Disord. 2016;18(4):356–66. https://doi.org/10.1684/epd.2016.0882 .

Lamberink HJ, Otte WM, Geerts AT, Pavlovic M, Ramos-Lizana J, Marson AG, et al. Individualised prediction model of seizure recurrence and long-term outcomes after withdrawal of antiepileptic drugs in seizure-free patients: a systematic review and individual participant data meta-analysis. Lancet Neurol. 2017;16(7):523–31. https://doi.org/10.1016/S1474-4422(17)30114-X .

Marson AG, Al-Kharusi AM, Alwaidh M, Appleton R, Baker GA, Chadwick DW, et al. The SANAD study of effectiveness of carbamazepine, gabapentin, lamotrigine, oxcarbazepine, or topiramate for treatment of partial epilepsy: an unblinded randomised controlled trial. Lancet. 2007;369(9566):1000–15. https://doi.org/10.1016/S0140-6736(07)60460-7 .

Marson AG, Al-Kharusi AM, Alwaidh M, Appleton R, Baker GA, Chadwick DW, et al. The SANAD study of effectiveness of valproate, lamotrigine, or topiramate for generalised and unclassifiable epilepsy: an unblinded randomised controlled trial. Lancet. 2007b;369(9566):1016–26. https://doi.org/10.1016/S0140-6736(07)60461-9 .

Marson A, Burnside G, Appleton R, Smith D, Leach JP, Sills G, et al. The SANAD II study of the effectiveness and cost-effectiveness of levetiracetam, zonisamide, or lamotrigine for newly diagnosed focal epilepsy: an open-label, non-inferiority, multicentre, phase 4, randomised controlled trial. Lancet. 2021;397(10282):1363–74. https://doi.org/10.1016/S0140-6736(21)00247-6 .

Marson A, Burnside G, Appleton R, Smith D, Leach JP, Sills G, et al. The SANAD II study of the effectiveness and cost-effectiveness of valproate versus levetiracetam for newly diagnosed generalised and unclassifiable epilepsy: an open-label, non-inferiority, multicentre, phase 4, randomised controlled trial. Lancet. 2021;397(10282):1375–86. https://doi.org/10.1016/S0140-6736(21)00246-4 .

Gesche J, Khanevski M, Solberg C, Beier CP. Resistance to valproic acid as predictor of treatment resistance in genetic generalized epilepsies. Epilepsia. 2017;58(4):e64–9. https://doi.org/10.1111/epi.13702 .

Meador K. Teratogenicity and antiseizure medications. Epilepsy Curr. 2020;20(6_suppl):15S–7S.

Article Google Scholar

Tomson T, Battino D, Perucca E. Teratogenicity of antiepileptic drugs. Curr Opin Neurol. 2019;32(2):246–52. https://doi.org/10.1097/WCO.0000000000000659 .

Tomson T, Battino D, Bonizzoni E, Craig J, Lindhout D, Perucca E, et al. Comparative risk of major congenital malformations with eight different antiepileptic drugs: a prospective cohort study of the EURAP registry. Lancet Neurol. 2018;17(6):530–8. https://doi.org/10.1016/S1474-4422(18)30107-8 .

Tomson T, Battino D, Bonizzoni E, Craig J, Lindhout D, Perucca E, et al. Declining malformation rates with changed antiepileptic drug prescribing: an observational study. Neurology. 2019;93(9):e831–40. https://doi.org/10.1212/WNL.0000000000008001 .

Steinhoff BJ, Staack AM. Levetiracetam and brivaracetam: a review of evidence from clinical trials and clinical experience. Ther Adv Neurol Disord. 2019;12:1756286419873518.

Subramonian A, Farrah K. Brivaracetam versus levetiracetam for epilepsy: a review of comparative clinical safety: Canadian Agency for Drugs and Technologies in Health, Ottawa, Ontario, Canada; 2020. Available at: https://cadth.ca/sites/default/files/pdf/htis/2020/RC1322%20Comparing%202%20Antiepileptic%20Drugs%20Final.pdf (accessed June 8, 2021)

Löscher W, Sills GJ, White HS. The ups and downs of alkyl-carbamates in epilepsy therapy: how does cenobamate differ? Epilepsia. 2021;62(3):596–614. https://doi.org/10.1111/epi.16832 .

Roberti R, De Caro C, Iannone LF, Zaccara G, Lattanzi S, Russo E. Pharmacology of cenobamate: Mechanism of action, pharmacokinetics, drug-drug interactions and tolerability. CNS Drugs. 2021. https://doi.org/10.1007/s40263-021-00819-8 (online ahead of print).

Krauss GL, Klein P, Brandt C, Lee SK, Milanov I, Milovanovic M, et al. Safety and efficacy of adjunctive cenobamate (YKP3089) in patients with uncontrolled focal seizures: a multicentre, double-blind, randomised, placebo-controlled, dose-response trial. Lancet Neurol. 2020;19(1):38–48. https://doi.org/10.1016/S1474-4422(19)30399-0 .

Vossler DG. Remarkably high efficacy of cenobamate in adults with focal-onset seizures: a double-blind, randomized, placebo-controlled trial. Epilepsy Curr. 2020;20(2):85–7. https://doi.org/10.1177/1535759720903032 .

Article PubMed PubMed Central Google Scholar

Franco V, Bialer M, Perucca E. Cannabidiol in the treatment of epilepsy: current evidence and perspectives for further research. Neuropharmacology. 2021;185:108442. https://doi.org/10.1016/j.neuropharm.2020.108442 .

Bialer M, Perucca E. Does cannabidiol have antiseizure activity independent of its interactions with clobazam? An appraisal of the evidence from randomized controlled trials. Epilepsia. 2020;61(6):1082–9. https://doi.org/10.1111/epi.16542 .

Ceulemans B, Schoonjans AS, Marchau F, Paelinck BP, Lagae L. Five-year extended follow-up status of 10 patients with Dravet syndrome treated with fenfluramine. Epilepsia. 2016;57(7):e129–34. https://doi.org/10.1111/epi.13407 .

Odi R, Invernizzi RW, Gallily T, Bialer M, Perucca E. Fenfluramine repurposing from weight loss to epilepsy: what we do and do not know. Pharmacol Ther. 2021;226:107866. https://doi.org/10.1016/j.pharmthera.2021.107866 .

Bialer M, Johannessen SI, Koepp MJ, Levy RH, Perucca E, Perucca P, et al. Progress report on new antiepileptic drugs: a summary of the fifteenth Eilat conference on new antiepileptic drugs and devices (EILAT XV). II. Drugs in more advanced clinical development. Epilepsia. 2020;61(11):2365–85. https://doi.org/10.1111/epi.16726 .

Overwater IE, Rietman AB, van Eeghen AM, de Wit MCY. Everolimus for the treatment of refractory seizures associated with tuberous sclerosis complex (TSC): current perspectives. Ther Clin Risk Manag. 2019;15:951–5. https://doi.org/10.2147/TCRM.S145630 .

French JA, Lawson JA, Yapici Z, Ikeda H, Polster T, Nabbout R, et al. 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(10056):2153–63. https://doi.org/10.1016/S0140-6736(16)31419-2 .

Fedak Romanowski EM, McNamara NA, Neil EE, Gottlieb-Smith R, Dang LT. Seizure rescue medications for out-of-hospital use in children. J Pediatr. 2021;229:19–25. https://doi.org/10.1016/j.jpeds.2020.10.041 .

Bouw MR, Chung SS, Gidal B, King A, Tomasovic J, Wheless JW, et al. Clinical pharmacokinetic and pharmacodynamic profile of midazolam nasal spray. Epilepsy Res. 2021;171:106567. https://doi.org/10.1016/j.eplepsyres.2021.106567 .

Boddu SHS, Kumari S. A short review on the intranasal delivery of diazepam for treating acute repetitive seizures. Pharmaceutics. 2020;12(12):1167. https://doi.org/10.3390/pharmaceutics12121167 .

Article CAS Google Scholar

Gidal B, Klein P, Hirsch LJ. Seizure clusters, rescue treatments, seizure action plans: unmet needs and emerging formulations. Epilepsy Behav. 2020;112:107391. https://doi.org/10.1016/j.yebeh.2020.107391 .

Fesler JR, Stanton S, Merner K, Ross L, McGinley MP, Bena J, et al. Bridging the gap in epilepsy care: a single-center experience of 3700 outpatient tele-epilepsy visits. Epilepsia. 2020;61(8):e95–e100. https://doi.org/10.1111/epi.16619 .

Alzamanan MZ, Lim KS, Akmar Ismail M, Abdul GN. Self-management apps for people with epilepsy: systematic analysis. JMIR Mhealth Uhealth. 2021;9(5):e22489. https://doi.org/10.2196/22489 .

Patterson V. The development of a smartphone application to help manage epilepsy in resource-limited settings. Seizure. 2020;79:69–74. https://doi.org/10.1016/j.seizure.2020.03.020 .

Giuliano L, Cicero CE, Trimarchi G, Todaro V, Colli C, Crespo Gómez EB, et al. Usefulness of a smartphone application for the diagnosis of epilepsy: validation study in high-income and rural low-income countries. Epilepsy Behav. 2021;115:107680. https://doi.org/10.1016/j.yebeh.2020.107680 .

Asadi-Pooya AA, Beniczky S, Rubboli G, Sperling MR, Rampp S, Perucca E. A pragmatic algorithm to select appropriate antiseizure medications in patients with epilepsy. Epilepsia. 2020;61(8):1668–77. https://doi.org/10.1111/epi.16610 .

Beniczky S, Rampp S, Asadi-Pooya AA, Rubboli G, Perucca E, Sperling MR. Optimal choice of antiseizure medication: agreement among experts and validation of a web-based decision support application. Epilepsia. 2021;62(1):220–7. https://doi.org/10.1111/epi.16763 .

Hadady L, Klivényi P, Asadi-Pooya AA, Rampp S, Fabó D, Bereczki C, et al. Web-based decision support system for patient-tailored selection of antiseizure treatment medication in adolescents and adults: An external validation study. 2021 (submitted).

Google Scholar

de Jong J, Cutcutache I, Page M, Elmoufti S, Dilley C, Fröhlich H, et al. Towards realizing the vision of precision medicine: AI based prediction of clinical drug response. Brain. 2021:awab108. https://doi.org/10.1093/brain/awab108 Online ahead of print.

Beniczky S, Wiebe S, Jeppesen J, Tatum WO, Brazdil M, Wang Y, et al. Automated seizure detection using wearable devices: a clinical practice guideline of the international league against epilepsy and the International Federation of Clinical Neurophysiology. Epilepsia. 2021;62(3):632–46. https://doi.org/10.1111/epi.16818 .

Rasheed K, Qayyum A, Qadir J, Sivathamboo S, Kwan P, Kuhlmann L, et al. Machine learning for predicting epileptic seizures using EEG signals: a review. IEEE Rev Biomed Eng. 2021;14:139–55. https://doi.org/10.1109/RBME.2020.3008792 .

Perucca P, Bahlo M, Berkovic SF. The genetics of epilepsy. Annu Rev Genomics Hum Genet. 2020;21(1):205–30. https://doi.org/10.1146/annurev-genom-120219-074937 .

Daci A, Bozalija A, Jashari F, Krasniqi S. Individualizing treatment approaches for epileptic patients with glucose transporter Type1 (GLUT-1) deficiency. Int J Mol Sci. 2018;19(1):122. https://doi.org/10.3390/ijms19010122 .

Article CAS PubMed Central Google Scholar

Perucca P, Perucca E. Identifying mutations in epilepsy genes: impact on treatment selection. Epilepsy Res. 2019;152:18–30. https://doi.org/10.1016/j.eplepsyres.2019.03.001 .

Specchio N, Pietrafusa N, Perucca E, Cross JH. New paradigms for the treatment of pediatric monogenic epilepsies: progressing toward precision medicine. Epilepsy Behav. 2021:107961. https://doi.org/10.1016/j.yebeh.2021.107961 Online ahead of print.

Byrne S, Enright N, Delanty N. Precision therapy in the genetic epilepsies of childhood. Dev Med Child Neurol. 2021. https://doi.org/10.1111/dmcn.14929 Online ahead of print.

Atkin TA, Maher CM, Gerlach AC, Gay BC, Antonio BM, Santos SC, et al. A comprehensive approach to identifying repurposed drugs to treat SCN8A epilepsy. Epilepsia. 2018;59(4):802–13. https://doi.org/10.1111/epi.14037 .

Löscher W, Potschka H, Sisodiya SM, Vezzani A. Drug resistance in epilepsy: clinical impact, potential mechanisms, and new innovative treatment options. Pharmacol Rev. 2020;72(3):606–38. https://doi.org/10.1124/pr.120.019539 .

Peng J, Pang N, Wang Y, Wang X-L, Chen J, Xiong J. Next-generation sequencing improves treatment efficacy and reduces hospitalization in children with drug-resistant epilepsy. CNS Neurosci Ther. 2019;25(1):14–20. https://doi.org/10.1111/cns.12869 .

Balestrini S, Chiarello D, Gogou M, Silvennoinen K, Puvirajasinghe C, Jones WD, et al. Real-life survey of pitfalls and successes of precision medicine in genetic epilepsies. J Neurol Neurosurg Psychiatry. 2021:jnnp-2020-325932. https://doi.org/10.1136/jnnp-2020-325932 Online ahead of print.

Kobylarek D, Iwanowski P, Lewandowska Z, Limphaibool N, Szafranek S, Labrzycka A, et al. Advances in the potential biomarkers of epilepsy. Front Neurol. 2019;10:685. https://doi.org/10.3389/fneur.2019.00685 .

Engel J Jr, Pitkänen A. Biomarkers for epileptogenesis and its treatment. Neuropharmacology. 2020;167:107735. https://doi.org/10.1016/j.neuropharm.2019.107735 .

Chang CJ, Chen CB, Hung SI, Ji C, Chung WH. Pharmacogenetic testing for prevention of severe cutaneous adverse drug reactions. Front Pharmacol. 2020;11:969. https://doi.org/10.3389/fphar.2020.00969 .

Simonato M, Agoston DV, Brooks-Kayal A, Dulla C, Fureman B, Henshall DC, et al. Identification of clinically relevant biomarkers of epileptogenesis - a strategic roadmap. Nat Rev Neurol. 2021;17(4):231–42. https://doi.org/10.1038/s41582-021-00461-4 .

Galanopoulou AS, Löscher W, Lubbers L, O'Brien TJ, Staley K, Vezzani A, et al. Antiepileptogenesis and disease modification: Progress, challenges, and the path forward-report of the preclinical working group of the 2018 NINDS-sponsored antiepileptogenesis and disease modification workshop. Epilepsia Open. 2021;6(2):276–96. https://doi.org/10.1002/epi4.12490 .

Kearney H, Byrne S, Cavalleri GL, Delanty N. Tackling epilepsy with high-definition precision medicine: a review. JAMA Neurol. 2019;76(9):1109–16. https://doi.org/10.1001/jamaneurol.2019.2384 .

Klein P, Friedman A, Hameed MQ, Kaminski RM, Bar-Klein G, Klitgaard H, et al. Repurposed molecules for antiepileptogenesis: missing an opportunity to prevent epilepsy? Epilepsia. 2020;61(3):359–86. https://doi.org/10.1111/epi.16450 .

Han Z, Chen C, Christiansen A, Ji S, Lin Q, Anumonwo C, et al. Antisense oligonucleotides increase Scn1a expression and reduce seizures and SUDEP incidence in a mouse model of Dravet syndrome. Sci Transl Med. 2020;12(558):eaaz6100.

Ahonen S, Nitschke S, Grossman TR, Kordasiewicz H, Wang P, Zhao X, et al. Gys1 antisense therapy rescues neuropathological bases of murine Lafora disease. Brain. 2021:awab194. https://doi.org/10.1093/brain/awab194 Online ahead of print.

Higurashi N, Broccoli V, Hirose S. Genetics and gene therapy in Dravet syndrome. Epilepsy Behav. 2021:108043. https://doi.org/10.1016/j.yebeh.2021.108043 Online ahead of print.

Cook M, Murphy M, Bulluss K, D'Souza W, Plummer C, Priest E, et al. Anti-seizure therapy with a long-term, implanted intra-cerebroventricular delivery system for drug-resistant epilepsy: a first-in-man study. EClinicalMedicine. 2020;22:100326. https://doi.org/10.1016/j.eclinm.2020.100326 .

Löscher W. The holy grail of epilepsy prevention: preclinical approaches to antiepileptogenic treatments. Neuropharmacology. 2020;167:107605. https://doi.org/10.1016/j.neuropharm.2019.04.011 .

Rahman MH, Akter R, Kamal MA. Prospective function of different antioxidant containing natural products in the treatment of neurodegenerative disease. CNS Neurol Disord drug targets 2020 Jul 22. Doi: https://doi.org/10.2174/1871527319666200722153611 . Online ahead of print.

Leavy A, Jimenez Mateos EM. Perinatal brain injury and inflammation: lessons from experimental murine models. Cells. 2020;9(12):2640. https://doi.org/10.3390/cells9122640 .

Yang N, Guan QW, Chen FH, Xia QX, Yin XX, Zhou HH, et al. Antioxidants targeting mitochondrial oxidative stress: promising neuroprotectants for epilepsy. Oxidative Med Cell Longev. 2020:6687185. https://doi.org/10.1155/2020/6687185 .

Zavala-Tecuapetla C, Cuellar-Herrera M, Luna-Munguia H. Insights into potential targets for therapeutic intervention in epilepsy. Int J Mol Sci. 2020;21(22):8573. https://doi.org/10.3390/ijms21228573 .

Frajewicki A, Laštůvka Z, Borbélyová V, Khan S, Jandová K, Janišová K, et al. Perinatal hypoxic-ischemic damage: review of the current treatment possibilities. Physiol Res. 2020;69(Suppl 3):S379–401. https://doi.org/10.33549/physiolres.934595 .

Murugan M, Boison D. Ketogenic diet, neuroprotection, and antiepileptogenesis. Epilepsy Res. 2020;167:106444. https://doi.org/10.1016/j.eplepsyres.2020.106444 .

Stone NL, Murphy AJ, England TJ, O'Sullivan SE. A systematic review of minor phytocannabinoids with promising neuroprotective potential. Br J Pharmacol. 2020;177(19):4330–52. https://doi.org/10.1111/bph.15185 .

Sanz P, Serratosa JM, Sánchez MP. Beneficial effects of metformin on the central nervous system, with a focus on epilepsy and Lafora disease. Int J Mol Sci. 2021;22(10):5351. https://doi.org/10.3390/ijms22105351 .

Tesfaye BA, Hailu HG, Zewdie KA, Ayza MA, Berhe DF. Montelukast: the new therapeutic option for the treatment of epilepsy. J Exp Pharmacol. 2021;13:23–31. https://doi.org/10.2147/JEP.S277720 .

Kotulska K, Kwiatkowski DJ, Curatolo P, Weschke B, Riney K, Jansen F, et al. Prevention of epilepsy in infants with tuberous sclerosis complex in the EPISTOP trial. Ann Neurol. 2021;89(2):304–14. https://doi.org/10.1002/ana.25956 .

Franco V, French JA, Perucca E. Challenges in the clinical development of new antiepileptic drugs. Pharmacol Res. 2016;103:95–104. https://doi.org/10.1016/j.phrs.2015.11.007 .

Download references

Acknowledgments

This article was not supported by any funding source.

Author information

Authors and affiliations.

Department of Internal Medicine and Therapeutics, Division of Clinical and Experimental Pharmacology, University of Pavia, Via Ferrata 9, 27100, Pavia, Italy

Emilio Perucca

Department of Neuroscience, Central Clinical School, Monash University, Melbourne, Victoria, 3004, Australia

You can also search for this author in PubMed Google Scholar

Contributions

The author was solely responsible for the conception and preparation of this article. The author read and approved the final manuscript.

Corresponding author

Correspondence to Emilio Perucca .

Ethics declarations

Ethics approval and consent to participate, consent for publication.

The author gave consent for publication of this article.

Competing interests

The author received speaker’s or consultancy fees from Angelini-Arvelle, Biogen, Eisai, GW Pharma, Sanofi, Sun Pharma, UCB Pharma, Xenon Pharma and Zogenix.

Additional information

This article is dedicated to the memory of Professor Alan Richens (1938–2021), my mentor and dear friend, in recognition of his outstanding contribution to advancing the pharmacological treatment of epilepsy.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and permissions

About this article

Cite this article.

Perucca, E. The pharmacological treatment of epilepsy: recent advances and future perspectives. Acta Epileptologica 3 , 22 (2021). https://doi.org/10.1186/s42494-021-00055-z

Download citation

Received : 28 June 2021

Accepted : 30 July 2021

Published : 17 September 2021

DOI : https://doi.org/10.1186/s42494-021-00055-z

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

Drug therapy
Antiepileptic drugs
Antiseizure medications

Acta Epileptologica

ISSN: 2524-4434

Submission enquiries: Access here and click Contact Us

2024 June 30
Epilepsy is one of the common neurological diseases which affects 65-70 million people worldwide. Modified Atkins diet (MAD) as a therapy is used as one of the treatments to reduce the seizures occurrence in epileptic patients. The purpose of this purpose is to review all eviden...

ORIGINAL ARTICLE 2024 June 30
Background and Purpose: Sumatriptan protects the brain from damage and enhance the anti-seizure effect of morphine. There is evidence that nitric oxide (NO) may mediate these effects of both drugs. In the present study, we investigated the effects of sumatriptan (0.1-20 mg/kg, i...

ORIGINAL ARTICLE 2024 June 30
Background and Purpose: Medication errors are common in the inpatient setting. Epilepsy patients who miss doses of their antiseizure medications are at risk for breakthrough seizures and subsequent complications. The purpose of this study was to quantify and characterize anti-se...

If anyone registers his/her email address, he or she will receive the email of notification when new paper is published in the web site of JER.
Please .

Review

Original Articles

Case Reports

--> !-->

To submit a manuscript, please visit the e-submission management system at , read the , and log into the e-submission system. For assistance with manuscript submission, please contact: .

Anyone may access any past or current articles without logging in.

KCC Parktown 101-308, Mallijae-ro 185, Jung-Gu, Seoul, Korea
TEL : +82-362-9662 FAX : +82-2-362-9663 E-mail :

Copyright© Korean Epilepsy Society.

| | |

An official website of the United States government

The .gov means it’s official. Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Publications
Account settings
My Bibliography
Collections
Citation manager

Save citation to file

Email citation, add to collections.

Create a new collection
Add to an existing collection

Add to My Bibliography

Your saved search, create a file for external citation management software, your rss feed.

Search in PubMed
Search in NLM Catalog
Add to Search

Revisiting the role of neurotransmitters in epilepsy: An updated review

Affiliations.

1 Faculty of Medicine, Department of Biophysics, Yozgat Bozok University, 66100 Yozgat, Turkey.
2 Medical Faculty, Yozgat Bozok University, 66100 Yozgat, Turkey.
3 Department of Biological Chemistry, Medical School, National and Kapodistrian University of Athens, Athens, Greece. Electronic address: [email protected].
4 Neuropharmacology Research Strength, Jeffrey Cheah School of Medicine and Health Sciences, Monash University Malaysia, Bandar Sunway, Selangor, Malaysia. Electronic address: [email protected].
PMID: 33259863
DOI: 10.1016/j.lfs.2020.118826

Epilepsy is a neurologicaldisorder characterized by persistent predisposition to recurrent seizurescaused by abnormal neuronal activity in the brain. Epileptic seizures maydevelop due to a relative imbalance of excitatory and inhibitory neurotransmitters. Expressional alterations of receptors and ion channelsactivated by neurotransmitters can lead to epilepsy pathogenesis.

Aims: In this updated comprehensive review, we discuss the emerging implication of mutations in neurotransmitter-mediated receptors and ion channels. We aim to provide critical findings of the current literature about the role of neurotransmitters in epilepsy.

Materials and methods: A comprehensive literature review was conducted to identify and critically evaluate studies analyzing the possible relationship between epilepsy and neurotransmitters. The PubMed database was searched for related research articles.

Key findings: Glutamate and gamma-aminobutyric acid (GABA) are the main neurotransmitters playing a critical role in the pathophysiology of this balance, and irreversible neuronal damage may occur as a result of abnormal changes in these molecules. Acetylcholine (ACh), the main stimulant of the autonomic nervous system, mediates signal transmission through cholinergic and nicotinic receptors. Accumulating evidence indicates that dysfunction of nicotinic ACh receptors, which are widely expressed in hippocampal and cortical neurons, may be significantly implicated in the pathogenesis of epilepsy. The dopamine-norepinephrine-epinephrine cycle activates hormonal and neuronal pathways; serotonin, norepinephrine, histamine, and melatonin can act as both hormones and neurotransmitters. Recent reports have demonstrated that nitric oxide mediates cognitive and memory-related functions via stimulating neuronal transmission.

Significance: The elucidation of the role of the main mediators and receptors in epilepsy is crucial for developing new diagnostic and therapeutic approaches.

Keywords: Epilepsy; Excitation; Inhibition; Ion channel; Neurotransmitters; Receptor.

PubMed Disclaimer

Publication types

Search in MeSH

Related information

PubChem Compound (MeSH Keyword)

LinkOut - more resources

Full text sources.

Elsevier Science
Ovid Technologies, Inc.

Other Literature Sources

scite Smart Citations
Genetic Alliance
MedlinePlus Consumer Health Information
MedlinePlus Health Information
Citation Manager

NCBI Literature Resources

MeSH PMC Bookshelf Disclaimer

The PubMed wordmark and PubMed logo are registered trademarks of the U.S. Department of Health and Human Services (HHS). Unauthorized use of these marks is strictly prohibited.

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

View all journals
Explore content
About the journal
Publish with us
Sign up for alerts
Open access
Published: 19 August 2024

Alterations in serum metabolomics during the first seizure and after effective control of epilepsy

Xiaolei Lian 1 , 3 na1 ,
Zhenguo Liu 2 , 4 na1 ,
Shanshuo Liu 2 , 4 na1 ,
Limin Jin 1 ,
Tianwen Wu 1 ,
Yuan Chen 1 ,
Shuang Li 1 ,
Wenzhong Kang 1 ,
Yajun Lian 1 ,
Yan Jiang 1 &
Zhigang Ren 2 , 4

Scientific Reports volume 14 , Article number: 19180 ( 2024 ) Cite this article

178 Accesses

Metrics details

Medical research
Molecular biology

The existing diagnostic methods of epilepsy have great limitations, and more reliable and less difficult diagnostic methods are needed. We collected serum samples of adult patients with first-diagnosed epilepsy (EPs) and seizure control patients (EPRs) for non-targeted metabolomics detection and found that they were both significantly altered, with increased expression of nicotine addiction, linoleic acid metabolism, purine metabolism, and other metabolic pathways. The diagnostic model based on 4 differential metabolites achieved a diagnostic efficiency of 99.4% in the training cohort and 100% in the validation cohort. In addition, the association analysis of oral flora, serum metabolism, and clinical indicators also provided a new angle to analyze the mechanism of epilepsy. In conclusion, this study characterized the serum metabolic characteristics of EPs and EPRs and the changes before and after epilepsy control based on a large cohort, demonstrating the potential of metabolites as non-invasive diagnostic tools for epilepsy.

NMR-based metabolomics in pediatric drug resistant epilepsy – preliminary results

Identification of cerebrospinal fluid and serum metabolomic biomarkers in first episode psychosis patients

Targeted metabolomics reveals aberrant profiles of serum bile acids in patients with schizophrenia

Introduction.

Epilepsy is one of the most common brain diseases, which is characterized by repetitive, episodic, and transient central nervous system dysfunction caused by excessive discharge of brain neurons 1 , 2 . Epilepsy affects more than 70 million people worldwide and affects people of all ages with high mortality and disability rates. The World Health Organization has listed it as one of the neuropsychiatric diseases that need to be treated and prevented urgently 3 , 4 . The diagnosis of epilepsy requires comprehensive judgment based on the patient's medical history, clinical manifestations and signs, electroencephalogram, neuroimaging, and genetic testing. Because of the complex etiology and seizure types of epilepsy, the choice of antiepileptic drugs is also very difficult. How to diagnose and treat epilepsy more timely and accurately has always been a hot spot and major difficulty in the field of international epilepsy research.

Metabolomics plays an important role in the study of systems biology, which has the advantages of high throughput, high sensitivity, and high precision, and can combine the changes of endogenous metabolites of organisms with the physiological and pathological changes of diseases. In recent years, metabolomics research has developed rapidly, and it is often used in the field of medical research to screen potential biomarkers of diseases, type diagnosis, prognosis, and efficacy evaluation. Metabolites are the result of local or systemic cellular responses 5 . Because metabolites can cross the blood–brain barrier, they may to some extent indicate changes in the biology of the central nervous system 6 , 7 . Beamer et al . found that blood levels of adenosine and its breakdown products can indicate epilepsy 8 . Engelke et al. 9 combined metabolomics with infrared ion spectroscopy and found that 6-oxoPIP can be used as a biomarker for the diagnosis of pyridoxine-dependent epilepsy (PDE-ALDH7A1). However, the metabolome of large epilepsy cohorts has not been characterized or studied.

Therefore, this study intends to reveal the changes in serum metabolomics of epileptic patients through a large cohort, clarify the material basis of epilepsy from the perspective of metabolomics, provide the biological basis and new ideas for further studies, and establish a diagnostic model by identifying differential potential biomarkers. In addition, we combined oral microbiome, serum metabolome, and clinical indicators to provide a broad theoretical basis for the early diagnosis and pathogenesis of epilepsy.

The research protocol and flow chart

A total of 944 samples from Henan Province were prospectively included in this study for statistical analysis, including 509 tongue swabs (186 EPs, 22 EPRs, 301 HCs) and 435 serum samples (131 EPs, 22 EPRs, 282 HCs). We randomly selected the serum samples of 100 EPs and 200 HCs as a discovery cohort, identified different metabolites and their key metabolic pathways between the two groups, obtained the best potential biomarkers using random forest and fivefold cross-validation, and established a diagnostic model. Furthermore, we verified the diagnostic capability of the classifier with serum samples of 31 EPs and 82 HCs. We performed an association analysis of tongue flora, serum metabolites, and laboratory indicators for the same individual (100 EPs and 200 HCs). We also explored the serum metabolomics of 22 EPRs and 44 HCs (19 EPs and 19 EPRs). The specific research framework is shown in Fig. 1 .

Research program. After strict inclusion and exclusion of 439 serum samples collected from Henan Province, 435 serum samples (282 HCs, 131 EPs, 22 EPRs) were analyzed. They were randomly divided into the discovery cohort and validation cohort, the former characterized metabolomic features, searched for markers, and established the prediction model by random forest method, while the latter conducted validation. In addition, Spearman correlation analysis was performed to explore the relationship between oral flora, serum metabolism, and clinical indicators among different populations. HCs, healthy controls; EPs, patients diagnosed with epilepsy; EPRs, patients whose seizures were under control; RFC, random forest classifier.

Basic information and clinical characteristics of participants

The basic information of 100 EPs and 200 HCs for serum samples was shown in Supplementary Table 1 . There was no significant difference in sex or age between EPs and HCs ( P > 0.05). Eps exhibited statistically significant differences in red blood cell (RBC) count, white blood cell (WBC) count, hemoglobin (Hb), albumin (ALB), uric acid (UA), estimated glomerular filtration rate (eGFR), and total bilirubin (TBIL) compared with HCs ( P < 0.05), but these values were all within the normal range.

Characterization of serum metabolomics of EPs and establishment of diagnostic model

We explored serum metabolomic characteristics of EPs and HCs using untargeted metabolomic approaches. The principal component analysis (PCA) score graph (Supplementary Fig. 1 A) showed a high degree of aggregation of quality control (QC), indicating good repeatability of QC and stability of the analysis system. In the discovery cohort, orthogonal partial least squares discrimination analysis (OPLS-DA) in multivariate statistical analysis showed that EPs and HCs were significantly separated (Fig. 2 A). The intercept of Q2 in the permutation test was less than 0, indicating that the fitting effect of the OPLS-DA model was very good (Fig. 2 B). Based on the identification and quantification of 1090 metabolites, 257 (23.58%) with intergroup differences were screened (variable importance in projection (VIP) > 1, P < 0.05) (Supplementary Table 2 ).

Characterization of serum metabolomics in EPs and HCs and the establishment of the diagnostic model. ( A ) The OPLS-DA score showed that the metabolites of the two groups (EPs, n = 100; HCs, n = 200) were significantly different. ( B ) The permutation test indicates that the model had a good fitting effect. ( C ) In the volcano plot, the dots on the left are differentially downregulated metabolites, and the dots on the right are differentially upregulated metabolites. The abscissa is the multiple of metabolite expression between the two groups (log2-fold change); the ordinate is the statistical test value of the difference in metabolite expression levels (-log10 ( P value )). ( D ) A KEGG topology bubble diagram shows the relative influence of different metabolites on metabolic pathways between the two groups. Each bubble in the figure represents a KEGG pathway, and the horizontal axis represents the relative importance of metabolites in the pathway. The vertical axis shows the enrichment significance of metabolite-involved pathways. Bubble size is represented by Impact Value. The color represents the P value for pathway enrichment. ( E ) The POD value of EPs was significantly higher than that of HCs in the discovery cohort. ( F ) In the discovery cohort, the AUC was 99.4%. ( G ) In the validation cohort, the POD value of EPs (n = 31) was significantly higher than that of HCs (n = 82). ( H ) In the validation cohort, the AUC is 100%. OPLS-DA, Orthogonal Partial Least Squares Discrimination Analysis; KEGG, Kyoto Encyclopedia of Genes and Genomes; POD, probability of disease; AUC, area under the ROC curve; HCs, healthy controls; EPs, patients diagnosed with epilepsy. Centerline, median; box limits, upper and lower quartiles; circle or square symbol, mean; error bars, 95% CI.

According to multiples of changes in metabolite levels in EPs relative to HCs, a volcano plot (Fig. 2 C) illustrated upregulated and downregulated metabolites. According to the heatmap (Supplementary Fig. 1 F, Supplementary Table 2 ), expression levels of 117 metabolites, including malathion monocarboxylic acid, serylproline, aspartyl-threonine, and 7-Methyl-3-oxo-6-octenoyl-CoA were increased in EPs, whereas levels of 140 metabolites, such as Asp-Phe, phenylalanylphenylalanine, and artonol B decreased. Moreover, we found that these elevated differential metabolites were mainly fatty acyls and carboxylic acids and derivatives. A line chart depicts the expression trend of the two subclusters in all samples (Supplementary Fig. 1 B), and the blue line represents the average expression of all metabolites in the subcluster.

KEGG (Kyoto Encyclopedia of Genes and Genomes) pathway enrichment analysis (Supplementary Fig. 1 D, Supplementary Table 3 ) was performed for different metabolites between groups. Of the 69 metabolic pathways that were enriched, 15 were statistically significant ( P < 0.05). Among them, Nicotine addiction, Caffeine metabolism, Choline metabolism in cancer, and other metabolic pathways were significantly different. Furthermore, we analyzed the relative effects of different metabolites on metabolic pathways, and D-glutamine and D-glutamate metabolism, cutin, suberine and wax biosynthesis, caffeine metabolism, lysine biosynthesis, and alanine, aspartate and glutamate metabolism were the most affected (Fig. 2 D).

In addition, we found four metabolite molecules that could most accurately distinguish epilepsy patients from healthy controls, and they were L-alpha-glutamyl-L-hydroxyproline, Vulgarone A, MG(0:0/14:1(9Z)/0:0) and Cis-3-Hexenyl phenylacetate, respectively (Supplementary Fig. 1 C, E, Supplementary Table 4 ). Based on the optimal tag set, the POD (Probability of Disease) index of EPs in the queue was significantly higher than that of HCs (Fig. 2 E), and the AUC (Area Under the Curve) reached 99.4% (95% CI: 98.24% to 100%, P < 0.0001) (Fig. 2 F). Furthermore, 31 EPs and 82 HCs were used as the validation cohort to test the diagnostic capability of the model. The results showed that the POD index of EP was also higher than that of HC (Fig. 2 G), with an AUC of 1 (95% CI: 100% to 100%, P < 0.0001) (Fig. 2 H). These results indicate that specific serum metabolites can be used to diagnose epilepsy.

Association analysis of oral microecology, serum metabolome, and laboratory indicators in Eps and HCs

We matched tongue swabs and serum samples from the same individual and obtained 100 EPs and 200 HCs. In association analysis of oral microflora, metabolites, and clinical indicators (Fig. 3 ), there was a significant correlation among 21 OTUs (Operational Taxonomic Units), 10 metabolites, and 6 clinical indicators. In correlation analysis between clinical indicators and metabolomics (Supplementary Fig. 2 A), ALB and TBIL correlated negatively with 7 metabolites (e.g., 7-methyl-3-oxo-6-octenoyl-CoA, malathion monocarboxylic acid) and positively with 3 metabolites (artonol B, Asp-Phe, and 10,11-dihydro-20-trihydroxy-leukotriene B4). In association analysis between clinical indicators and microbe species (Supplementary Fig. 2 B, 21 OTUs (e.g., OTU113 ( Prevotella nigrescens ) and OTU22 ( Lautropia mirabilis )) correlated negatively with RBCs, ALB, and TBIL. Spearman correlation analysis (Supplementary Fig. 2 C) between 67 OTUs and 14 metabolites showed that 58 OTUs (e.g., OTU90 ( Granulicatella elegans ), OTU162 ( Porphyromonas catoniae ), and OTU85 ( Corynebacterium matruchotii )) correlated positively with 11 metabolites (e.g., 10-hydroxycarbazepine, portulacaxanthin III, polyporusterone B, aspartyl-threonine) and negatively with Asp-Phe, artonol B and 10,11-dihydro-20-trihydroxy-leukotriene B4. In contrast, 5 OTUs, OTU12 ( Peptostreptococcus stomatis ) and OTU21 ( Solobacterium moorei ), correlated negatively with the above 11 metabolites and positively with the above three metabolites. These results suggest the correlation between oral microecology, serum metabolomics, and some clinical indicators.

Association analysis of the oral microecology, serum metabolome, and laboratory indicators in Eps. There are significant correlations among 6 laboratory indicators, 21 OTUs, and 10 metabolites. Red lines indicate negative correlations, blue lines indicate positive correlations, and the width of the lines represents the strength of the correlation (Spearman). The transparency of the lines represented the negative logarithm of the P-value of correlation, translucent lines meant (− lg P ) > 5, and opaque lines meant (− lg P ) > 10. The size of the points indicates the relative abundance of genera and metabolites. The colors of points display the different phyla of the microbiome. The circle represents the oral microbiome, the square represents the laboratory indicators, and the diamond represents metabolites. OTUs operational taxonomy units; WBC, white blood cells; RBC, red blood cells; PLT, platelet; ALB, albumin; UA, uric acid; TBIL, total bilirubin; B, bacteria; PI, phenotype index; Meta, metabolite; rho, correlation index.

Characterization of the serum metabolites in EPRs and HCs

We followed previously enrolled EPs and eventually collected tongue swabs and serum samples from 22 recovered patients (EPRs). The serum of 22 EPRs and 44 HCs was analyzed by untargeted metabolomics. Under the condition that the analysis system is stable (Fig. 4 A) and the fitting effect of the OPLS-DA model is good (Fig. 4 C), OPLS-DA can well distinguish between EPRs and HCs (Fig. 4 B). 229 metabolites with intergroup differences were selected from 1095 identified and quantified metabolites (VIP > 1, P < 0.05) (Supplementary Table 5 ). We used a volcano plot (Fig. 4 E) to illustrate metabolites with upregulated and downregulated expression.

Metabolomic characteristics of EPRs (n = 22) are different from those of HCs (n = 44). ( A ) PCA score diagram results showed that the QC polymerization degree was high, and the analysis system was stable. ( B ) OPLS-DA showed a significant dispersion of metabolites between the two groups. ( C ) The permutation test indicates that the model had a good fitting effect. ( D ) The expression trend line chart showed the changes in the expression levels of metabolites in each group in each sample. The blue line represents the average expression levels of all metabolites in the subcluster. ( E ) The volcano plot shows downregulated and upregulated metabolites. ( F ) A KEGG topology bubble diagram shows the relative influence of different metabolites on metabolic pathways between the two groups. ( G ) Enrichment analysis histogram shows the enriched metabolic pathways of differential metabolites. *, p < 0.05, **, p < 0.01, ***, p < 0.001. PCA, principal component analysis; QC, quality control; OPLS-DA, Orthogonal Partial Least Squares Discrimination Analysis; KEGG, Kyoto Encyclopedia of Genes and Genomes; HCs, healthy controls; EPRs, patients whose seizures are under control.

The heatmap shows the relative expression levels of key metabolites in each sample. Compared with HCs, relative expression levels of 91 metabolites (e.g., malathion monocarboxylic acid, 7-Methyl-3-oxo-6-octenoyl-CoA, and aspartyl-threonine) in EPRs increased, and relative expression levels of 138 metabolites decreased. Notably, many of these differential metabolites are consistent with the differential metabolites found in EPs and HCs comparisons. The variation trends of EPRs and HCs metabolites in the two kinds of metabolite clusters were not consistent (Fig. 4 D). In the 86 pathways of KEGG enrichment analysis (Fig. 4 G, Supplementary Table 6 ), 18 metabolic pathways had statistical significance. Among them, nicotine addiction, leishmaniasis, caffeine metabolism, sphingolipid metabolism, and other metabolic pathways were significantly different. According to metabolic pathway analysis, D-glutamine and D-glutamate metabolism and sphingolipid metabolism were most affected between the two groups (Fig. 4 F).

Three metabolite molecules including Serylisoleucine, 1-Methyladenosine and Heptaethylene glycol were identified by random forest analysis and fivefold cross-validation (Supplementary Table 7 ). Based on the optimal tag set, the POD index of EPRs was significantly higher than that of HCs, with an AUC reaching 100% (95% CI 100% to 100%, P < 0.0001).

Correlation analysis of the oral microecology, metabolic spectrum, and clinical indicators in EPRs and HCs

We performed an association analysis of metabolomics, oral microecology, and laboratory indicators for 22 EPRs and 44 HCs. As shown in Supplementary Fig. 3 D, 10 OTUs, 10 metabolites, and TBIL showed a strong correlation. Heptathylene glycol correlated negatively with 5 OTUs (e.g., OTU30 ( Lautropia mirabilis ), OTU34 ( Rothia aeria ), OTU78 ( Actinomyces HMT 169 )). OTU30 ( Lautropia mirabilis ), OTU34 ( Rothia aeria ), OTU78 ( Actinomyces HMT 169 ), and OTU133 ( Porphyromonas catoniae ) correlated positively with 9 metabolites but not heptethylene. We further depict the correlation between any two of the metabolites, microbial species, and laboratory indicators in Supplementary Fig. 3 A, B, and C.

Characterization of serum metabolites before and after seizure control

To determine changes in metabolites and oral microflora before and after seizure control, we characterized the oral microecology and metabolomics of 19 EPs and 19 EPRs. LC–MS was performed on 19 EPs and serum samples. The results of the PCA score chart (Supplementary Fig. 4 A) showed that the degree of polymerization of QC was high and that the data quality was reliable. In the case of the good fitting effect of the OPLS-DA model (Fig. 5 B), EPs and EPRs were discrete (Fig. 5 A). Furthermore, 18 intergroup differential metabolites were screened out from 1095 identified and quantified metabolite molecules (VIP > 1, P < 0.05) (Supplementary Table 8 ).

Metabolomics differences before and after seizure control. ( A ) The OPLS-DA score showed obvious metabolite dispersion of EPs (n = 19) and EPRs (n = 19). ( B ) The permutation test indicates that the model had a good fitting effect. ( C ) The volcano plot shows downregulated and upregulated metabolites. ( D ) A KEGG topology bubble diagram shows the relative influence of different metabolites on metabolic pathways between the two groups. ( E ) The heatmap shows the relative expression of each metabolite in all samples. ( F ) 12 metabolites were selected as the best potential biomarkers. ( G ) In the discovery cohort, the POD value of EPRs was significantly higher than that of EPs. ( H ) In the discovery cohort, the AUC was 1. OPLS-DA, Orthogonal Partial Least Squares Discrimination Analysis; KEGG, Kyoto Encyclopedia of Genes and Genomes; POD, probability of disease; AUC, area under the ROC curve; EPs, patients diagnosed with epilepsy; EPRs, patients whose seizures were under control. Centerline, median; box limits, upper and lower quartiles; circle or square symbol, mean; error bars, 95% CI.

A volcano plot (Fig. 5 C) shows multiples of differences in metabolites between the two groups. According to the heatmap (Fig. 5 E), the levels of 15 metabolic molecules in EPRs increased compared with EPs, whereas the levels of 3 metabolic molecules decreased. We observed that the average levels of metabolites in EPs and EPRs were different (Supplementary Fig. 4 B). Among the 42 KEGG pathways to which different metabolites were enriched, 13 had statistical significance (Supplementary Fig. 4 C, Supplementary Table 9 ). There were significant differences in metabolic pathways such as sphingolipid metabolism, and choline metabolism in cancer and necroptosis. Furthermore, MetaboAnalystR (Fig. 5 D) suggested that different metabolites had the most significant effect on sphingolipid metabolism, followed by vitamin B6 metabolism, tyrosine metabolism, toluene degradation, and glycerophospholipid metabolism.

Finally, we performed a quintupled cross-validation analysis of differential metabolites to obtain a marker composed of 12 metabolites (Fig. 5 F and Supplementary Fig. 4 D, Supplementary Table 10 ). Based on the metabolite marker set, the POD index of EPRs in this cohort was significantly higher than that of EPs (Fig. 5 G), and the AUC reached 100% (95% CI 100% to 100%, P < 0.0001) (Fig. 5 H).

Correlation analysis of the oral microecology, metabolic spectrum, and clinical indicators before and after seizure control

We performed an association analysis of the microecology, metabolic spectrum, and clinical indicators in 19 EPRs and 19 EPs (Supplementary Fig. 5 ). In correlation analysis of 10 OTUs and 2 metabolic molecules (Supplementary Fig. 4 F), (3-(2,5-dihydroxyphenyl)-3-oxo-1-phenylpropoxy) sulfonic acid correlated negatively with 6 OTUs (e.g., OTU17 ( Veillonella atypica ), OTU20 ( Prevotella pallens )) but positively with OTU55 ( Neisseria oralis ). RBCs correlated positively with 10 OTUs (e.g., OTU51 ( Neisseria elongata ) and OTU8 ( Neisseria perflava )) (Supplementary Fig. 4 G). N-Acetyl-dl-tryptophan correlated negatively with Cr and ALB and BUN positively with (3-(2,5-dihydroxyphenyl)-3-oxo-1-phenylpropoxy) sulfonic acid (Supplementary Fig. 4 E).

The occurrence of epilepsy requires the participation of genetic, physiological, and environmental factors, and its repeated characteristics will not only bring physical and psychological obstacles to patients but also cause serious social and economic burdens. At present, the lack of effective early diagnostic tools makes the intervention treatment and prognosis recovery of epilepsy difficult 10 . The search for biomarkers that can be used in clinical diagnosis and tracking of disease-causing processes will help advance patient care, disease research, and the birth of new treatment strategies. Among the different biological samples, cerebrospinal fluid is the most accurate for diagnosis of epilepsy, but blood not only provides more comprehensive information, but the sampling is minimally invasive 11 . As a very ideal clinical diagnostic medium, its small molecule metabolites can cross the blood–brain barrier and may well represent the biological phenomena occurring in the central nervous system. Metabolomics identifies disease biomarkers in an efficient and high-throughput manner, elucidates complex biological mechanisms, monitors therapeutic effects, and is considered an effective tool in all aspects of disease diagnosis, classification, staging, treatment, and prognosis 12 , 13 . An in-depth understanding of metabolomic changes during the development of epilepsy can provide new insights into its pathophysiological mechanisms and provide new ideas for the development of effective anti-epilepsy strategies.

In this study, we compared the serum metabolites of a large sample of patients with epilepsy with those of healthy subjects and identified the serum metabolites in the former that differ from those in the latter. A total of 1090 metabolites were identified and quantified by LC–MS non-targeted metabolomics analysis, and 257 intergroup differential metabolites were found. Compared with HCs, Eps had 117 differently elevated metabolite levels, including malathion monocarboxylic acid, serylproline, aspartyl-threonine, and 7-Methyl-3-oxo-6-octenoyl-CoA, while 140 metabolites such as Asp—Phe, phenylalanylphenylalanine, artonol B, and 4—Hydroxy—alprenolol levels decreased. These elevated differential metabolites were mainly fatty acyls and carboxylic acids and derivatives. This is consistent with other studies, considering that it is caused by abnormal fatty acid metabolism during epilepsy 14 , 15 . Through further analysis of the KEGG pathway, the five most affected metabolic pathways were found, namely D-glutamine and D-glutamate metabolism, cutin, suberine and wax biosynthesis, caffeine metabolism, lysine biosynthesis, and alanine, aspartate and glutamate metabolism. By searching for differential metabolites and projecting them into the corresponding metabolic pathways, we found that adult patients with their first untriggered seizure had higher levels of caffeine metabolism, nicotine addiction, linoleic acid metabolism, choline metabolism in cancer, biosynthesis of amino acids, D-Glutamine and D-glutamate metabolism, purine metabolism, arginine biosynthesis, alanine, aspartate and glutamate metabolism and taste transduction compared with healthy volunteers, both at the time of illness and after 3–4 months of recovery. Brain imaging studies in humans have shown that nicotine can activate brain regions such as the prefrontal cortex, thalamus, and basal ganglia, and even had toxic effects at high doses that can cause nausea, confusion, seizures, and even death 16 . Nicotine addiction can enhance the ability of dopamine transmission, inducing long-term potentiation (LTP) of glutamatergic synapses in the ventral tegmental area (VTA) dopamine neurons 17 . As an excitatory neurotransmitter, glutamate had been shown to play an important role in epileptic seizures. In addition, the role played by purinergic signaling systems in mediating the excitability of neuronal networks was becoming increasingly clear 18 . It had also been confirmed that the top two enriched metabolic pathways involved in pediatric drug-resistant epilepsy are unsaturated fatty acid biosynthesis and linoleic acid metabolism 19 . Compared with EP, metabolites involved in sphingolipid metabolism in EPR were increased. Sphingolipid metabolism plays an important role in the homeostasis and function of the central nervous system and is closely related to Alzheimer's disease, Parkinson's disease, and multiple sclerosis 20 . Consistent with some previous findings, our study confirmed that serum metabolites were altered in adults with first unprovoked seizures and short-term control, and some metabolic pathways that enhance nervous system excitability or increase drug tolerance were increased. We believe that these are some key metabolic pathways in epileptogenesis and have certain significance for the diagnosis and treatment of epilepsy.

In addition, through the correlation analysis of oral microbiota, serum metabolomics, and laboratory indicators, we found that oral microbiota and serum metabolites were changed in patients with epilepsy and short-term control, and there was a correlation between oral microflora and serum metabolites, suggesting that oral microbiota and serum metabolomics may be linked in the regulation of epilepsy. The presence of metabolites and flora that remain at abnormal levels suggests that there may still be an inherent pathogenic biochemical basis for EPR. Metabolome-based markers have also been used in the diagnosis of diabetes 21 and nonalcoholic steatohepatitis 22 . We used random forest analysis to construct relevant potential biomarkers based on serum metabolites and validated the diagnostic efficacy of the model with a randomized cohort. This diagnostic model will be helpful for early screening or diagnosis of epilepsy, especially when the diagnosis of epilepsy is not clear, and it may be an important auxiliary means.

In conclusion, based on serum metabolomics analysis, this study determined the changes in serum metabolites in EPs, EPRs, and HCs, and Spearman correlation analysis was performed to find the relationship between oral flora, serum metabolites, and clinical indicators during the onset of epilepsy. The diagnostic model based on potential biomarkers achieved excellent diagnostic performance in both discovery and validation queues. Specific serum metabolites reflect the changes and characteristics of epilepsy and seizure control in different states of the disease, providing a large sample of reliable evidence for the study of serum metabolism of epilepsy and new ideas for the development of effective anti-epilepsy strategies.

Research program

This study was approved by the Institutional Review Board of the First Affiliated Hospital of Zhengzhou University (No. 2021-KY-0574-002). All samples and clinical data involved in this study were collected with the informed consent of each participant. A total of 131 serum samples were prospectively collected from hospitalized adult patients with the first unprovoked seizure of epilepsy (epileptic patients, EPs). After 3–4 months of follow-up, seizures were controlled in 22 patients without using antiepileptic drugs (epileptic patients recovered, EPRs). The healthy controls (HCs) were 527 volunteers who underwent a physical examination in the First Affiliated Hospital of Zhengzhou University. Serum samples were subjected to untargeted metabolomics analysis. Study inclusion and exclusion criteria and the collection and testing of tongue coating specimens were in Supplementary Methods.

Serum sample preparation

The anterior elbow venous blood was collected into a common serum tube (BD, Oxford, UK), and the supernatant was collected immediately after centrifugation at 4 °C and 10,000 rpm for 10 min. Stored the supernatant at − 80 °C. For testing, the sample was thawed at 4 °C and precisely removed 100 µL into a 1.5 ml centrifuge tube. Then 400 µL methanol containing 0.02 mg/mL internal standard (L-2-chlorophenylalanine) was added and swirled for 30 s. Low-temperature ultrasonic extraction for 30 min (temperature 5 °C, power 40 kHz). Then we placed the sample at − 20 °C for 30 min, centrifuged it at 4 °C at a speed of 13,000 g for 15 min, and transferred the supernatant to a liquid chromatograph-mass spectrometer (LC–MS) injection vial for machine testing. In addition, each sample was mixed with 20 µL supernatant and used as the quality control (QC) sample. A tube of 200 µL serum was taken out to test liver function, renal function, and other indicators in the laboratory department of the First Affiliated Hospital of Zhengzhou University.

LC–MS experiment

The LC–MS analysis was performed using a tandem ultra-high performance liquid chromatography-Fourier transform mass spectrometry system (UHPLC-Q Exactive HF-X, Thermo Fisher Scientific, USA). The chromatographic column was ACQUITY UPLC HSS T3 (100 mm × 2.1 mm i.d., 1.8 µm; Waters, Milford, USA), the mobile phase A was 95% water + 5% acetonitrile (containing 0.1% formic acid), and the mobile phase B was 47.5% acetonitrile + 47.5% isopropyl alcohol + 5% water (containing 0.1% formic acid). The sample size was 2 μL, and the column temperature was 40 °C. The samples were ionized by electrospray, and the mass spectrum signals were collected by positive and negative ion scanning modes respectively. The quality control samples were prepared by mixing the extraction liquid of all samples in the same volume, and the volume of each QC was the same as that of the sample, which was processed and tested in the same way as the analytical sample. In the process of instrument analysis, a QC sample was inserted into every 5–15 analytical samples to investigate the stability of the entire detection process.

Chromatogram of total ions

The components of the sample separated by chromatography continuously entered the mass spectrum, and the mass spectrum was continuously scanned for data collection. A mass spectrum was obtained for each scan, and the total ion current intensity was obtained by adding all the ion intensities in each mass spectrum. The total ion chromatograms that time as the abscissa, ionic strength sum as ordinate. The detection effect was evaluated according to the total ion chromatograms of quality control samples in positive and negative ion modes.

Identification of metabolites

The raw data were imported into Progenesis QI (Waters Corporation, Milford, USA) for baseline filtering, peak identification, integration, retention time correction, peak alignment, etc. Finally, the data matrix containing retention time, mass-to-charge ratio, and peak intensity information was obtained. The MS and MS/MS mass spectrum information were matched with the metabolic database. MS mass error was set to less than 10 ppm, and metabolites were identified according to the secondary mass spectrum matching score. The main database for metabolites identification were some mainstream public databases and self-built databases including Human Metabolome Database ( http://www.hmdb.ca/ ), MetaboAnalyst ( https://www.metaboanalyst.ca/ ) and Metabolite Link ( https://metlin.scripps.edu/ ).

Multivariate data analysis

After the data matrix was imported into the “ropls” package (v1.18.8) of R 4.1.1 ( http://www.R-project.org/ ) 23 , the unsupervised principal component analysis (PCA) method was used to analyze the data, which could be used to find abnormal samples and evaluate QC repeatability. A supervised (orthogonal) partial least squares analysis (OPLS-DA) was then used to show the overall differences in metabolic profiles among groups and to look for the metabolites that differed between groups. In the OPLS-DA analysis, variables whose Variable Importance in Projection (VIP) was greater than 1 were important variables. The corrected T-test and multivariate analysis of OPLS-DA were used to screen out the metabolites with differences between groups (VIP > 1 and P value < 0.05). To prevent the model from overfitting, the fitting effect of the model was investigated by using 200 substitution tests.

Advanced analysis

For the differential metabolites obtained through multivariate statistical analysis, MetaboAnalystR 4.0 ( https://github.com/xia-lab/MetaboAnalystR ) was used for metabolite aggregation analysis 24 , and then we performed pathway enrichment analysis of the differential metabolites using Metabolic Pathway Analysis (MetPA) module (KEGG PATHWAY Database, http://www.genome.jp/kegg/pathway.html ) 25 . To quantify pathway activity, the composite score for each pathway was standardized to 1, where the importance measure for each biomolecule was given a weighted score based on the relative importance of its position. Finally, the cumulative importance score of the current pathway was obtained by calculating the weighted score of corresponding metabolites. The higher the score, the greater the influence of the pathway. In addition, we used the“ggplot2” package (v3.3.5) in R 4.1.1 ( http://www.R-project.org/ ) to perform a clustering analysis of the expression patterns of metabolites with statistical differences (Metabolite distance algorithm: Euclidean; Metabolite level clustering method: Complete) 26 .

Construction of diagnostic models

Serum samples from epilepsy patients and healthy controls were randomly divided into training cohort and test cohort at a ratio of 2:1. In the training cohort, we characterized the serum metabolites of 100 EPs and 200 HCs, constructed a classifier based on the random forest model (randomForest package (v4.6.14)) 27 , and selected the best combination of biomarkers by five-fold cross validation. According to the determined optimal set of biomarkers, the probability of disease 12 index was calculated for the training cohort and the test cohort. The POD index is the ratio of the number of samples predicted to be EPs to the number of samples predicted to be HC in a randomly generated decision tree. Then, the receiver operating characteristic (ROC) curve was constructed using the “pROC” package (v1.17.0.1) of R 28 , and the performance of the model was evaluated by the area under the curve. In addition, correlations among oral microbiota, lipid molecules and clinical indicators were elucidated based on Spearman correlation analysis.

Statistical analysis

The continuous variables with normal distribution were expressed as mean ± standard deviation and compared between the two groups using the t-test. Continuous variables that were not normally distributed were expressed as medians and interquartile ranges, and comparisons between groups were performed with the use of the Wilcoxon rank-sum test. The chi-square test or Fisher's exact test was used for categorical variables between the two groups. The statistical analysis was performed using SPSS V.26 for Windows (SPSS, Chicago, Illinois, USA). Statistical significance was defined by P < 0.05.

Limitations of the study

This study included adults who had their first seizure without an apparent trigger and who did not have another seizure after a short follow-up, and the reason for inclusion was partly related to clinical limitations. However, the clinical classification of epilepsy was very complex, and we only considered one of them. Adding more epilepsy classifications for subgroup analysis or long-term follow-up will provide more evidence for the diagnosis and treatment of epilepsy. In addition, we only found some metabolites and metabolic pathways related to seizures, and more specific mechanisms have not been detected. In-depth verification may bring more surprises.

Ethics approval and consent to participate

This study was approved by the Institutional Review Board from the First Affiliated Hospital of Zhengzhou University (No. 2021-KY-0574-002). The study was performed in accordance with the Helsinki Declaration and Rules of Good Clinical Practice. All participants signed written informed consent after the study protocol was fully explained.

Patient and public involvement

Patients or the public were not involved in the design, conduct, reporting, or dissemination plans of our research.

Data availability

The raw Illumina read data for all tongue coating samples were deposited in the European Bioinformatics Institute European Nucleotide Archive database (Accession Number: PRJNA759716). The serum metabolomics data sets supporting the conclusions of this article are included within the article and its additional files. Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact: Zhigang Ren ([email protected]).

Sheng, J., Liu, S., Qin, H., Li, B. & Zhang, X. Drug-resistant epilepsy and surgery. Curr. Neuropharmacol. 16 , 17–28. https://doi.org/10.2174/1570159x15666170504123316 (2018).

Article CAS PubMed PubMed Central Google Scholar

Liu, X. & Chen, J. Research progress on ferroptosis and its role in epilepsy. J. Physiol. Pharmacol. Off. J. Pol. Physiol. Soc. https://doi.org/10.26402/jpp.2022.6.02 (2022).

Article Google Scholar

Thijs, R. D., Surges, R., O’Brien, T. J. & Sander, J. W. Epilepsy in adults. Lancet (London, England) 393 , 689–701. https://doi.org/10.1016/s0140-6736(18)32596-0 (2019).

Article PubMed Google Scholar

Devinsky, O. et al. Epilepsy. Nat. Rev. Dis. Primers 4 , 18024. https://doi.org/10.1038/nrdp.2018.24 (2018).

Goodacre, R., Vaidyanathan, S., Dunn, W. B., Harrigan, G. G. & Kell, D. B. Metabolomics by numbers: Acquiring and understanding global metabolite data. Trends Biotechnol. 22 , 245–252. https://doi.org/10.1016/j.tibtech.2004.03.007 (2004).

Article CAS PubMed Google Scholar

Liu, P. et al. Discovery of metabolite biomarkers for acute ischemic stroke progression. J. Proteome Res. 16 , 773–779. https://doi.org/10.1021/acs.jproteome.6b00779 (2017).

Article ADS CAS PubMed Google Scholar

Abela, L. et al. N(8)-acetylspermidine as a potential plasma biomarker for Snyder-Robinson syndrome identified by clinical metabolomics. J. Inherit. Metab. Dis. 39 , 131–137. https://doi.org/10.1007/s10545-015-9876-y (2016).

Beamer, E. et al. Elevated blood purine levels as a biomarker of seizures and epilepsy. Epilepsia 62 , 817–828. https://doi.org/10.1111/epi.16839 (2021).

Engelke, U. F. et al. Untargeted metabolomics and infrared ion spectroscopy identify biomarkers for pyridoxine-dependent epilepsy. J. Clin. Investing. https://doi.org/10.1172/jci148272 (2021).

Niu, D., Sun, P., Zhang, F. & Song, F. Metabonomic analysis of cerebrospinal fluid in epilepsy. Ann. Transl. Med. 10 , 449. https://doi.org/10.21037/atm-22-1219 (2022).

Donatti, A., Canto, A. M., Godoi, A. B., da Rosa, D. C. & Lopes-Cendes, I. Circulating metabolites as potential biomarkers for neurological disorders-metabolites in neurological disorders. Metabolites https://doi.org/10.3390/metabo10100389 (2020).

Article PubMed PubMed Central Google Scholar

Murgia, F. et al. Metabolomics as a tool for the characterization of drug-resistant epilepsy. Front. Neurol. 8 , 459. https://doi.org/10.3389/fneur.2017.00459 (2017).

Eid, T. Harnessing metabolomics to advance epilepsy research. Epilepsy Curr. 22 , 123–129. https://doi.org/10.1177/15357597221074518 (2022).

Abuknesha, N. R. et al. Plasma fatty acid abnormality in Sudanese drug-resistant epileptic patients. Prostaglandins Leukot. Essent. Fatty Acids 167 , 102271. https://doi.org/10.1016/j.plefa.2021.102271 (2021).

Taha, A. Y., Burnham, W. M. & Auvin, S. Polyunsaturated fatty acids and epilepsy. Epilepsia 51 , 1348–1358. https://doi.org/10.1111/j.1528-1167.2010.02654.x (2010).

Melis, M. & Pistis, M. Targeting the interaction between fatty acid ethanolamides and nicotinic receptors: Therapeutic perspectives. Pharmacol. Res. 86 , 42–49. https://doi.org/10.1016/j.phrs.2014.03.009 (2014).

Mansvelder, H. D. & McGehee, D. S. Long-term potentiation of excitatory inputs to brain reward areas by nicotine. Neuron 27 , 349–357. https://doi.org/10.1016/s0896-6273(00)00042-8 (2000).

Beamer, E., Kuchukulla, M., Boison, D. & Engel, T. ATP and adenosine-Two players in the control of seizures and epilepsy development. Prog. Neurobiol. 204 , 102105. https://doi.org/10.1016/j.pneurobio.2021.102105 (2021).

Guo, H. L. et al. Integrating metabolomics and lipidomics revealed a decrease in plasma fatty acids but an increase in triglycerides in children with drug-refractory epilepsy. Epilepsia open 8 , 466–478. https://doi.org/10.1002/epi4.12712 (2023).

Alaamery, M. et al. Role of sphingolipid metabolism in neurodegeneration. J. Neurochem. 158 , 25–35. https://doi.org/10.1111/jnc.15044 (2021).

Song, L. et al. Urine metabonomics reveals early biomarkers in diabetic cognitive dysfunction. J. Proteome Res. 16 , 3180–3189. https://doi.org/10.1021/acs.jproteome.7b00168 (2017).

Qi, S. et al. Metabonomics screening of serum identifies pyroglutamate as a diagnostic biomarker for nonalcoholic steatohepatitis. Clin. Chim. Acta Int. J. Clin. CHEM. 473 , 89–95. https://doi.org/10.1016/j.cca.2017.08.022 (2017).

Article CAS Google Scholar

Thévenot, E. A., Roux, A., Xu, Y., Ezan, E. & Junot, C. Analysis of the human adult urinary metabolome variations with age, body mass index, and gender by implementing a comprehensive workflow for univariate and OPLS statistical analyses. J. Proteome Res. 14 , 3322–3335. https://doi.org/10.1021/acs.jproteome.5b00354 (2015).

Chong, J., Wishart, D. S. & Xia, J. Using MetaboAnalyst 4.0 for comprehensive and integrative metabolomics data analysis. Curr. Protoc. Bioinf. 68 , 86. https://doi.org/10.1002/cpbi.86 (2019).

Lou, J. et al. Fecal microbiomes distinguish patients with autoimmune hepatitis from healthy individuals. Front. Cell. Infect. Microbial. 10 , 342. https://doi.org/10.3389/fcimb.2020.00342 (2020).

Coleman, A., Bose, A. & Mitra, S. Metagenomics data visualization using R. Methods Mol. Boil. (Clifton, N.J.) 2649 , 359–392. https://doi.org/10.1007/978-1-0716-3072-3_19 (2023).

Alderden, J. et al. Predicting pressure injury in critical care patients: A machine-learning model. Am. J. Crit. Care Off. Pub. Am. Assoc. Crit. Care Nurses 27 , 461–468. https://doi.org/10.4037/ajcc2018525 (2018).

Robin, X. et al. pROC: An open-source package for R and S+ to analyze and compare ROC curves. BMC Bioinf. 12 , 77. https://doi.org/10.1186/1471-2105-12-77 (2011).

Download references

Acknowledgements

We thank all the generous volunteer subjects who enrolled in the study. We are also very grateful to Hongyan Ren and Chao Liu (Shanghai Mobio Biomedical Technology Co., Ltd.) for their generous help in sequencing.

This study was sponsored by grants from Key Scientific Research Project of Henan Province University (24A320031), Key project of Henan Provincial Natural Science Foundation (HNSZRKXJJZDXM2023019), Central Plains Talent Program-Central Plains Youth Top Talents, Young and Middle-aged Academic Leaders of Henan Provincial Health Commission (HNSWJW-2022013), and Funding for Scientific Research and Innovation Team of The First Affiliated Hospital of Zhengzhou University (QNCXTD2023002).

Author information

These authors contributed equally: Xiaolei Lian, Zhenguo Liu, and Shanshuo Liu.

Authors and Affiliations

Department of Neurology, The First Affiliated Hospital of Zhengzhou University, #1 Jianshe East Road, Zhengzhou, 450052, China

Xiaolei Lian, Limin Jin, Tianwen Wu, Yuan Chen, Shuang Li, Wenzhong Kang, Yajun Lian & Yan Jiang

Department of Infectious Diseases, State Key Laboratory of Antiviral Drugs, Pingyuan Laboratory, The First Affiliated Hospital of Zhengzhou University, #1 Jianshe East Road, Zhengzhou, 450052, China

Zhenguo Liu, Shanshuo Liu & Zhigang Ren

The Academy of Medical Sciences of Zhengzhou University, Zhengzhou University, Zhengzhou, 450052, China

Xiaolei Lian

Gene Hospital of Henan Province; Precision Medicine Center, The First Affiliated Hospital of Zhengzhou University, Zhengzhou, 450052, China

You can also search for this author in PubMed Google Scholar

Contributions

Z.R. and Y.J. designed the study. X.L, Y.J., Y.L., Z.L., L.J., T.W., Y.C., and S.L. collected clinical samples. X.L., L.J., T.W., and W.K. collected and analyzed the clinical data of the subjects. Z.L. and S.L extracted the bacterial DNA and analyzed the data. Z.R. and Y.J. performed MiSeq sequencing. Z.L. and S.L. wrote the manuscript, Z.R. revised the manuscript. All authors reviewed and approved the manuscript.

Corresponding authors

Correspondence to Yan Jiang or Zhigang Ren .

Ethics declarations

Competing interests.

The authors declare no competing interests.

Additional information

Publisher's note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Supplementary information 1., supplementary information 2., supplementary information 3., supplementary information 4., supplementary information 5., supplementary information 6., supplementary information 7., supplementary information 8., supplementary information 9., supplementary information 10., rights and permissions.

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/ .

Reprints and permissions

About this article

Cite this article.

Lian, X., Liu, Z., Liu, S. et al. Alterations in serum metabolomics during the first seizure and after effective control of epilepsy. Sci Rep 14 , 19180 (2024). https://doi.org/10.1038/s41598-024-68966-8

Download citation

Received : 05 January 2024

Accepted : 30 July 2024

Published : 19 August 2024

DOI : https://doi.org/10.1038/s41598-024-68966-8

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

By submitting a comment you agree to abide by our Terms and Community Guidelines . If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Quick links

Explore articles by subject
Guide to authors
Editorial policies

Introduction
Conclusions
Article Information

RCT indicates randomized clinical trial.

D indicates domain.

The studies by Devinsky et al 76 , 77 and O’Brien et al 79 have a high risk of bias; the studies by Devinsky et al, 75 Ben-Menachem et al, 74 and VanLandingham et al 82 have some concerns; and the studies by Thiele et al 80 , 81 and Miller et al 78 have a low risk of bias. DL indicates DerSimonian and Laird.

eTable 1. Search Strategy Used With the Four Databases

eTable 2. Quality Assessment of the Included Studies

eTable 3. Baseline Characteristics of the Studies Included in the Meta-analysis

eTable 4. Concomitant Anti-epileptic Drugs Taken by Participants in the Included Trials, N (%)

eTable 5. Adverse Events Observed in the Included Trials, N (%)

eTable 6. Graded Adverse Events in the Meta-analysis

eFigure 1. Percentages of Any-Grade Adverse Events for the Cannabidiol and Control Groups

eFigure 2. Percentages of Mild, Moderate and Severe Adverse Events for the Cannabidiol and Control Groups

eFigure 3. Percentages of Adverse Events Leading to the Discontinuation of the Trial for the Cannabidiol and Control Groups

eFigure 4. Forest Plot of the Risk Ratios for Severe Grade Adverse Events for the Cannabidiol and Control Groups

eFigure 5. Forest Plots of the Risk Ratios for Any-Grade Adverse Events, Including ALT or AST Elevation, Decreased Appetite, Diarrhea, Fatigue, Nasopharyngitis, and Pneumonia for the Cannabidiol and Control Groups

eFigure 6. Forest Plots of the Risk Ratios for Any-Grade Adverse Events, Including Pyrexia, Rash, Somnolence, Status Epilepticus, Upper Respiratory Tract Infection, and Vomiting for the Cannabidiol and Control Groups

eFigure 7. Forest Plot of the Risk Ratios for Mild Adverse Events, Including Decreased Appetite, Diarrhea, Nasopharyngitis, Pyrexia, Somnolence, and Vomiting for the Cannabidiol and Control Groups

eFigure 8. Forest Plot of the Risk Ratios for Moderate Adverse Events, Including Decreased Appetite, Diarrhea, Pyrexia, Somnolence, and Vomiting for the Cannabidiol and Control Groups

eFigure 9. Forest Plot of the Risk Ratios for Severe Adverse Events, Including Decreased Appetite, Diarrhea, and Somnolence for the Cannabidiol and Control Groups

eFigure 10. Forest Plot of the Risk Ratio for Serious Adverse Events for the Cannabidiol and Control Groups

eFigure 11. Forest Plot of the Risk Ratio for Adverse Events Leading to the Discontinuation the Trial for the Cannabidiol and Control Groups

eFigure 12. Forest Plot of the Risk Ratios for Adverse Events Leading to the Discontinuation of the Trial, Including ALT or AST Elevation, Diarrhea, and Rash for the Cannabidiol and Control Groups

eFigure 13. Forest Plot of the Risk Ratios for Adverse Events Leading to Dose Reduction for the Cannabidiol and Control Groups

eFigure 14. Forest Plot of the Risk Ratio for Any-Grade Adverse Events for the Cannabidiol and Control Groups by Quality of the Included Studies

Data Sharing Statement

See More About

Customize your JAMA Network experience by selecting one or more topics from the list below.

Academic Medicine
Acid Base, Electrolytes, Fluids
Allergy and Clinical Immunology
American Indian or Alaska Natives
Anesthesiology
Anticoagulation
Art and Images in Psychiatry
Artificial Intelligence
Assisted Reproduction
Bleeding and Transfusion
Caring for the Critically Ill Patient
Challenges in Clinical Electrocardiography
Climate and Health
Climate Change
Clinical Challenge
Clinical Decision Support
Clinical Implications of Basic Neuroscience
Clinical Pharmacy and Pharmacology
Complementary and Alternative Medicine
Consensus Statements
Coronavirus (COVID-19)
Critical Care Medicine
Cultural Competency
Dental Medicine
Dermatology
Diabetes and Endocrinology
Diagnostic Test Interpretation
Drug Development
Electronic Health Records
Emergency Medicine
End of Life, Hospice, Palliative Care
Environmental Health
Equity, Diversity, and Inclusion
Facial Plastic Surgery
Gastroenterology and Hepatology
Genetics and Genomics
Genomics and Precision Health
Global Health
Guide to Statistics and Methods
Hair Disorders
Health Care Delivery Models
Health Care Economics, Insurance, Payment
Health Care Quality
Health Care Reform
Health Care Safety
Health Care Workforce
Health Disparities
Health Inequities
Health Policy
Health Systems Science
History of Medicine
Hypertension
Images in Neurology
Implementation Science
Infectious Diseases
Innovations in Health Care Delivery
JAMA Infographic
Law and Medicine
Leading Change
Less is More
LGBTQIA Medicine
Lifestyle Behaviors
Medical Coding
Medical Devices and Equipment
Medical Education
Medical Education and Training
Medical Journals and Publishing
Mobile Health and Telemedicine
Narrative Medicine
Neuroscience and Psychiatry
Notable Notes
Nutrition, Obesity, Exercise
Obstetrics and Gynecology
Occupational Health
Ophthalmology
Orthopedics
Otolaryngology
Pain Medicine
Palliative Care
Pathology and Laboratory Medicine
Patient Care
Patient Information
Performance Improvement
Performance Measures
Perioperative Care and Consultation
Pharmacoeconomics
Pharmacoepidemiology
Pharmacogenetics
Pharmacy and Clinical Pharmacology
Physical Medicine and Rehabilitation
Physical Therapy
Physician Leadership
Population Health
Primary Care
Professional Well-being
Professionalism
Psychiatry and Behavioral Health
Public Health
Pulmonary Medicine
Regulatory Agencies
Reproductive Health
Research, Methods, Statistics
Resuscitation
Rheumatology
Risk Management
Scientific Discovery and the Future of Medicine
Shared Decision Making and Communication
Sleep Medicine
Sports Medicine
Stem Cell Transplantation
Substance Use and Addiction Medicine
Surgical Innovation
Surgical Pearls
Teachable Moment
Technology and Finance
The Art of JAMA
The Arts and Medicine
The Rational Clinical Examination
Tobacco and e-Cigarettes
Translational Medicine
Trauma and Injury
Treatment Adherence
Ultrasonography
Users' Guide to the Medical Literature
Vaccination
Venous Thromboembolism
Veterans Health
Women's Health
Workflow and Process
Wound Care, Infection, Healing

Get the latest research based on your areas of interest.

Others also liked.

Download PDF
X Facebook More LinkedIn

Fazlollahi A , Zahmatyar M , ZareDini M, et al. Adverse Events of Cannabidiol Use in Patients With Epilepsy : A Systematic Review and Meta-analysis . JAMA Netw Open. 2023;6(4):e239126. doi:10.1001/jamanetworkopen.2023.9126

Manage citations:

Permissions

Adverse Events of Cannabidiol Use in Patients With Epilepsy : A Systematic Review and Meta-analysis

1 Student Research Committee, Tabriz University of Medical Sciences, Tabriz, Iran
2 Research Center for Integrative Medicine in Aging, Aging Research Institute, Tabriz University of Medical Sciences, Tabriz, Iran
3 Social Determinants of Health Research Center, Tabriz University of Medical Sciences, Tabriz, Iran
4 Systematic Review and Meta-analysis Expert Group, Universal Scientific Education and Research Network, Tehran, Iran
5 Department of Life and Health Sciences, University of Nicosia, Nicosia, Cyprus
6 Department of Social Sciences, University of Nicosia, Nicosia, Cyprus
7 Brain Mapping Research Center, Shahid Beheshti University of Medical Sciences, Tehran, Iran
8 Social Determinants of Health Research Center, Shahid Beheshti University of Medical Sciences, Tehran, Iran
9 Neurosciences Research Center, Aging Research Institute, Tabriz University of Medical Sciences, Tabriz, Iran
10 Department of Community Medicine, Faculty of Medicine, Tabriz University of Medical Sciences, Tabriz, Iran

Question What are the frequency and risk of adverse events (AEs) developing in patients with epilepsy who are using cannabidiol (CBD)?

Findings In this systematic review and meta-analysis, the frequency of any grade AEs in patients with epilepsy was more than 2 times higher for those using CBD than for those receiving placebo. The risks of any grade AEs, severe grade AEs, serious AEs, AEs leading to discontinuation, and AEs leading to dose reduction were significantly higher in patients receiving CBD than for those receiving placebo.

Meaning The treatment of patients with epilepsy using CBD was associated with the development of several types of AEs.

Importance Epilepsy is one of the most common neurologic disorders globally. Cannabidiol (CBD) has been approved for the treatment of epilepsy, but its use has been associated with several different adverse events (AEs).

Objective To investigate the frequency and risk of AEs developing in patients with epilepsy who are using CBD.

Data Sources PubMed, Scopus, Web of Science, and Google Scholar were searched for relevant studies published from database inception up to August 4, 2022. The search strategy included a combination of the following keywords: ( cannabidiol OR epidiolex ) AND ( epilepsy OR seizures ).

Study Selection The review included all randomized clinical trials that investigated at least 1 AE from the use of CBD in patients with epilepsy.

Data Extraction and Synthesis Basic information about each study was extracted. I 2 statistics were calculated using Q statistics to assess the statistical heterogeneity among the included studies. A random-effects model was used in cases of substantial heterogeneity, and a fixed-effects model was used if the I 2 statistic for the AEs was lower than 40%. This study was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) guideline.

Main Outcomes and Measures Frequency of each AE and risk of developing each AE in patients with epilepsy using CBD.

Results Nine studies were included. Overall incidences of 9.7% in the CBD group and 4.0% in the control group were found for any grade AEs. The overall risk ratios (RRs) for any grade and severe grade AEs were 1.12 (95% CI, 1.02-1.23) and 3.39 (95% CI, 1.42-8.09), respectively, for the CBD group compared with the control group. Compared with the control group, the CBD group had a greater risk for incidence of serious AEs (RR, 2.67; 95% CI, 1.83-3.88), AEs resulting in discontinuation (RR, 3.95; 95% CI, 1.86-8.37), and AEs resulting in dose reduction (RR, 9.87; 95% CI, 5.34-14.40). Because most of the included studies had some risk of bias (3 raised some concerns and 3 were at high risk of bias), these findings should be interpreted with some caution.

Conclusions and Relevance In this systematic review and meta-analysis of clinical trials, the use of CBD to treat patients with epilepsy was associated with an increased risk of several AEs. Additional studies are needed to determine the safe and effective CBD dosage for treating epilepsy.

Epilepsy is one of the most common neurologic disorders globally, with a lifetime point prevalence of 7.6 per 1000 population and an annual incidence of 67 per 100 000 population. 1 Although most cases can be treated or go into remission with age, in approximately one-third of cases the seizures continue despite pharmacotherapy, surgical, or dietary interventions. 2 - 6 Therefore, it is important to find new alternatives for treating epilepsy.

Cannabidiol (CBD) is one of the naturally occurring compounds, known as cannabinoids, that are produced from the cannabis plant. In contrast to Δ8-tetrahydrocannabinol, CBD is not intoxicating at typical doses and lacks euphoric and other psychotropic effects. 7 , 8 Cannabidiol is approved by both the US Food and Drug Administration and the European Medicines Agency as an additional therapy for severe forms of epilepsy, such as Dravet syndrome and Lennox-Gastaut syndrome. 9 , 10

Previous studies have evaluated the efficacy and safety of CBD, or medicinal cannabis, in patients with epilepsy. 11 , 12 Moreover, a study published in 2020 evaluated the adverse events (AEs) associated with CBD use across all medical indications, 13 but to the best of our knowledge, no systematic review has focused on the AEs associated with CBD use in patients with epilepsy. Moreover, there is a need to update the previously published systematic reviews and to address their limitations. Therefore, we conducted a systematic review and meta-analysis to evaluate the AEs associated with CBD use in patients with epilepsy.

For this systematic review and meta-analysis, PubMed, Scopus, and the Web of Science databases were searched for articles published from database inception to August 4, 2022, to identify publications reporting any AEs following treatment with CBD. In addition, the first 10 pages of the Google Scholar search engine were manually searched for grey literature. No filters were applied to any of the search fields, such as date, study type, or language. Backward and forward citation searches of all included studies were also performed, which means we screened all cited references of the included studies and all publications citing them to discover other qualified studies. In addition, studies included in similar previous systematic reviews were screened to identify whether there were any additional eligible articles. The searches were performed by 1 author (A.F.) and then double-checked by other authors (S.A.N. and S.S.). The search strategy included a combination of the following keywords: ( cannabidiol OR epidiolex ) AND ( epilepsy OR seizures ). A detailed description of the stages of the search for each database is given in eTable 1 in Supplement 1 . The current study was approved by the ethics committee of the Shahid Beheshti University of Medical Sciences. This study was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-analyses ( PRISMA ) guideline. 14

All articles identified through the electronic and manual searches were exported to EndNote, version 20 (Clarivate), and any duplicates were removed. Two authors (A.F. and M. ZareDini) independently screened the title and abstract of the articles and excluded those that were irrelevant. In the next step, the same 2 authors reviewed the full texts of the remaining articles. Any discrepancies were resolved by discussion or consultation with other authors.

Studies were included if they were randomized clinical trials (RCTs) investigating at least 1 AE associated with CBD use in patients with epilepsy. All classifications of epilepsy were included, with no age restriction. Studies were excluded if they were not RCTs, did not consider the AEs of CBD, or included patients with diseases other than epilepsy.

Data extraction was conducted using previously designed Microsoft Office Excel forms (Microsoft Corp). Two reviewers (M. Zahmatyar and B.G.) independently extracted the following information from each included study: (1) the basic information about the study, including title, first author’s name, country, and publication date; (2) the characteristics of the participants, including study population, sample size, age, sex, type of epilepsy, and medications used for the treatment of epilepsy; and (3) the total number and severity of all-cause and treatment-related AEs observed in both the experimental and control groups as well as the total number of AEs resulting in discontinuation or dose reduction in both groups. Any disagreements were settled through discussion between the 2 reviewers or by conferring with a third reviewer (A.F.). Negative clinical events that developed in study participants after administration of CBD or placebo were considered to be AEs. We categorized the AEs according to the Common Terminology Criteria for Adverse Events, version 5.0. 15

Two reviewers (M. Zahmatyar and B.G.) independently appraised the risk of bias and the quality of the included articles using version 2 of the Cochrane risk-of-bias (RoB2) tool for randomized trials. 16 The RoB2 tool classifies studies as having a high, low, or unclear risk of bias (some concerns) using the following 5 domains: randomization process, deviations from the intended interventions, missing outcome data, measurement of the outcome, and selection of the reported results. The overall risk-of-bias assessment in each study was also determined. Disagreements were resolved by discussion between the 2 reviewers (M. Zahmatyar and B.G.) or by consulting with a third reviewer (A.F.). We used the robvis package in R software, version 4.2.2 (R Foundation for Statistical Computing) to create the risk-of-bias graph. 17

Stata, version 17.0 (StataCorp LLC) was used to perform the meta-analysis. We determined the frequency of mild, moderate, severe, and all grade AEs in both experimental and control groups using the metaprop command in Stata. 18 The dichotomous raw data on the frequency of AEs in the intervention and control arms were extracted from each included study. We calculated the I 2 statistics using Q statistics to assess the statistical heterogeneity among the included studies. The DerSimonian and Laird method was used for the random-effects models, and the inverse variance method was used for the fixed-effect models. A random-effects model was used in case of substantial heterogeneity, and a fixed-effects model was used if the I 2 statistic was lower than 40% for the AEs. 19 , 20 Continuity correction of 0.5 was used when the number of AEs for at least 1 arm was zero. Publication bias was only evaluated if at least 10 studies were included in the analysis. 21 P < .05 was considered statistically significant.

The systematic search identified 3280 records. After the removal of 1350 duplicate records, the remaining 1930 publications were screened, and 61 studies were selected for full-text review ( Figure 1 ). Following the evaluation of these studies for eligibility, 52 were excluded for the following reasons: 48 studies were not RCTs, 22 - 69 2 did not report AEs, 70 , 71 and 2 were reanalyses of previously published articles. 72 , 73 Finally, a total of 9 articles met the eligibility criteria and were included in the qualitative and quantitative synthesis. 74 - 82

Bias due to randomization, deviation from the intended intervention, and missing data were low risk in all included trials. However, bias due to outcome measurement and the selection and reporting of results were high risk or there were some concerns in several studies. Overall, 3 trials had a low risk of bias, 78 , 80 , 81 3 had some concerns, 74 , 75 , 82 and 3 had a high risk of bias 76 , 77 , 79 ( Figure 2 ; eTable 2 in Supplement 1 ).

The included trials were published between 2017 and 2022 and involved patients with various forms of epilepsy (Dravet syndrome, Lennox-Gastaut syndrome, and tuberous sclerosis–associated epilepsy). One study 79 used 390- and 195-mg transdermal CBD gels twice a day, whereas other studies 74 - 78 , 80 - 82 used oral solutions twice daily. The daily oral dose of CBD ranged from 5 to 50 mg/kg, and the duration of treatment ranged from 3 to 16 weeks. Most of the studies were multicenter, and the age of participants ranged from 1.1 to 56.8 years ( Table 1 ). The numbers of previous and concomitant antiepileptic drugs were fairly similar between the experimental and control arms (eTable 3 in Supplement 1 ). The concomitant antiepileptic drugs included valproate, clobazam, lamotrigine, levetiracetam, rufinamide, vigabatrin, stiripentol, lacosamide, ethosuximide, topiramate, zonisamide, oxcarbazepine, carbamazepine, lorazepam, clonazepam, eslicarbazepine, perampanel, and phenobarbital (eTable 4 in Supplement 1 ).

The number and percentage of AEs reported in each included study can be found in eTable 5 in Supplement 1 . In the intervention group, the most common AE of any grade was somnolence (22.0%), followed by a decreased appetite (19.5%) and pyrexia (15.3%). In the controls, upper respiratory tract infection (11.8%), diarrhea (10.9%), and pyrexia (10.2%) were the most common AEs. The overall percentage of any grade AEs was higher in the CBD group than in the control group (9.7% vs 4.0%) (eFigure 1 in Supplement 1 ). In the CBD arm, the overall percentages were 11.1% for mild AEs, 3.1% for moderate AEs, and 1.2% for severe AEs. In the control arm, the overall percentages were 6.4% for mild AEs, 1.3% for moderate AEs, and 0.7% for severe AEs (eFigure 2 in Supplement 1 ). The percentage of AEs that led to the discontinuation of the trial was higher in the CBD arm than in the controls (2.4% vs 0.7%) (eFigure 3 in Supplement 1 ).

The overall risk ratios (RRs) of any grade (from all 9 studies) and severe grade (in 5 studies 74 , 75 , 79 , 81 , 82 ) AEs in the CBD group compared with the control group were 1.12 (95% CI, 1.02-1.23; I 2 = 58.9%) for any grade and 3.39 (95% CI, 1.42-8.09; I 2 = 3.5%) for severe grade ( Figure 3 ; eFigure 4 in Supplement 1 ). For any grade AEs, the incidences of diarrhea (RR, 1.93; 95% CI, 1.44-2.58; I 2 = 0.0%), somnolence (RR, 2.29; 95% CI, 1.61-3.25; I 2 = 0.0%), decreased appetite (RR, 2.13; 95% CI, 1.48-3.06; I 2 = 10.2%), and alanine transaminase (ALT) or aspartate aminotransferase (AST) elevation (12.29; 95% CI, 4.22-35.80; I 2 = 0.0%) were significantly higher in the CBD group ( Table 2 ; eFigures 5 and 6 in Supplement 1 ).

Among mild grade AEs, the RR for the incidence of diarrhea was significantly higher in the CBD group (RR, 1.71; 95% CI, 1.21-2.42; I 2 = 0.0%) (eFigure 7 in Supplement 1 ). Among moderate grade AEs, the risks of decreased appetite (RR, 3.25; 95% CI, 1.20-8.83; I 2 = 0.0%) and somnolence (RR, 3.62; 95% CI, 1.45-9.04; I 2 = 0.0%) were significantly higher in those receiving CBD (eFigure 8 in Supplement 1 ). There was no significant difference between the CBD group and controls in terms of risk of severe grade site-specific AEs (eFigure 9 and eTable 6 in Supplement 1 ).

Using data from 8 of the included studies, 74 - 81 the overall RR for the incidence of serious AEs in the CBD group compared with the control group was 2.67 (95% CI, 1.83-3.88; I 2 = 8.9%). Furthermore, in 2 studies, the CBD group had a higher RR for serious AEs (RR, 6.68; 95% CI, 1.63-27.38 80 and RR, 4.94; 95% CI, 1.76-13.85) 81 (eFigure 10 in Supplement 1 ).

The incidence of AEs leading to the discontinuation of treatment in 8 studies 74 - 77 , 79 - 82 were significantly higher in the CBD group than in the control group (RR, 3.95; 95% CI, 1.86-8.37; I 2 = 0.0%) (eFigure 11 in Supplement 1 ). However, there were no significant differences between the CBD and control groups for AST or ALT elevation, diarrhea, and rashes leading to discontinuation of the trial (eFigure 12 in Supplement 1 ). The incidence of AEs that resulted in dose reduction in the 3 included studies 75 , 81 , 82 was significantly higher in the CBD group than in the control group (RR, 9.87; 95% CI, 5.34-14.40; I 2 = 0.0%) (eFigure 13 in Supplement 1 ).

A subgroup analysis was performed by quality of the included studies. The results showed that the pooled RRs for incidence of overall any grade AEs were 1.05 (95% CI, 0.96-1.16; I 2 = 62.9%) in studies with low risk of bias, 1.15 (95% CI, 0.95-1.39; I 2 = 33.6%) in studies with high risk of bias, and 1.35 (95% CI, 0.95-1.92; I 2 = 12.8%) in studies with some concerns (eFigure 14 in Supplement 1 ).

The current study showed that the frequency of any grade AEs in patients with epilepsy was more than 2 times higher for those receiving CBD than for the controls, with a notably increased risk of ALT or AST elevation, decreased appetite, diarrhea, and somnolence in those receiving CBD. The risks of any grade AEs, severe grade AEs, serious AEs, AEs leading to the discontinuation of the trial, and AEs leading to dose reduction were significantly higher in patients receiving CBD than in controls.

The current study also indicated that for those receiving CBD, the overall percentage of any grade AEs was 9.7%, severe grade AEs was 1.2%, and AEs leading to the discontinuation of the trial was 2.4%. A previous systematic review and meta-analysis 11 of RCTs involving patients with uncontrolled epilepsy showed that the frequency of any grade AEs was 55.7% and the frequency of serious AEs was 17.6%. The lower percentage of AE frequency found in our study may be as a result of including more studies in our analysis (9 vs 4). 11 In addition, differences in the inclusion criteria in the systematic reviews and definitions of AEs can lead to variations in the AE incidence. Moreover, a systematic review by Bilbao and Spanagel 83 on the safety and efficacy of cannabinoids for different diseases showed that serious or severe AEs occurred in 4.5% of those receiving CBD, which was lower than for other cannabinoids, such as dronabinol (5.4%) and nabilone (6.3%). Therefore, it seems that CBD is a relatively safe option compared with other cannabinoids.

The current study showed that use of CBD was associated with a 1.2 times increase in the incidence of any grade AEs, 3.39 times increase in the incidence of severe grade AEs, 2.67 times increase in the incidence of serious AEs, 3.95 times increase in the incidence of AEs leading to discontinuation of the trial, and 9.87 times increase in the incidence of AEs resulting in dose reduction compared with controls. Similarly, a systematic review by Lattanzi et al 11 found an RR of 1.22 (95% CI, 1.11-1.33) for the incidence of AEs in those receiving CBD compared with controls. Moreover, another meta-analysis 84 of 3 trials on the safety of adjunctive CBD in patients with Dravet syndrome showed that adding CBD was nonsignificantly associated with an increased risk of developing any type of AE (RR, 1.06; 95% CI, 0.87-1.28). Their nonsignificant result may be because they only included Dravet syndrome, had a low sample size, and included articles in the analysis that evaluated the effects of CBD that was administered in combination with other drugs. 84 Another meta-analysis 85 of 2 trials with 396 participants showed a significantly higher RR for AE development in patients with Lennox-Gastaut syndrome being treated with CBD (RR, 1.24; 95% CI, 1.11-1.38). Furthermore, the results of a systematic review and meta-analysis 86 on the efficacy and safety of CBD for pediatric refractory epilepsy found an increased risk of overall AEs of 1.81 (95% CI, 1.33-2.46) and of serious AEs of 2.86 (95% CI, 1.63-5.05). The risk of any grade and serious AEs in our study were lower than the above-mentioned study (1.12 vs 1.81 for any grade AEs and 2.67 vs 2.86 for serious AEs), 86 which may be because of the inclusion of different populations and AE definitions.

The current study found that there was a significantly increased incidence of diarrhea, somnolence, decreased appetite, and ALT or AST elevation among those receiving CBD. Similarly, 3 previous systematic reviews and meta-analyses 11 , 84 , 85 of patients with uncontrolled epilepsy (ie, Dravet syndrome and Lennox-Gastaut syndrome) showed that CBD was significantly associated with increased risks of somnolence, decreased appetite, diarrhea, and increased serum aminotransferases. Although Ben-Menachem et al 74 did not show notable increases in the RRs for ALT or AST elevation in their study, in most trials, CBD was associated with an increased RR of ALT or AST elevation when compared with the placebo group. These differences may be due to the fact that the studies were examining the effect of CBD treatment after a different period (25 days vs 14 weeks) in low numbers of patients with wide age ranges. 75 , 80 , 81 Thiele et al 80 found a significant RR for elevated ALT or AST levels as a result of treatment with high-dose CBD (25 mg/kg daily and 50 mg/kg daily) using a long treatment period compared with other studies (16 vs 14 weeks).

Among mild, moderate, and severe manifestations of somnolence, the moderate type was statistically significantly higher than in the control group in analyses of the pooled data from 4 studies 75 , 78 , 80 , 81 (RR, 3.62; 95% CI, 1.45-9.04). In most of these studies, taking clobazam was associated with somnolence. 75 , 78 , 80 , 81 For instance, in 1 of the RCTs, 75 approximately 80% of the patients with reported somnolence were also taking clobazam. In another study, 80 41% of the CBD group taking clobazam concomitantly reported somnolence as an AE in contrast to 12.5% among those not taking clobazam. Drug-drug interactions between clobazam and CBD were associated with an almost 3-fold increase in exposure to clobazam’s active metabolite, N-desmethyl clobazam, and an approximately 1.5-fold increase in CBD’s active metabolite, 7-hydroxy cannabidiol, in healthy volunteers. However, several studies involving patients with epilepsy did not demonstrate any drug-drug interaction between CBD and clobazam, whereas exposure to N-desmethyl clobazam was enhanced by 2- to 3-fold, probably resulting from the inhibition of CYP2C19 by CBD, which leads to clobazam dose adjustments in the presence of CBD. 78 , 81 , 82 , 87

A previous systematic review and meta-analysis of RCTs, 13 which used the RoB1 tool for quality assessment, showed that in most domains, most studies had low risks of bias, and only 2 studies had high risks of bias in the selective outcome reporting domains. Similarly, in our study, which used the RoB2 tool, we found that there were high risks of bias in the selection of reported results and in the measurement of outcomes, whereas other domains had low risks of bias. The study by Talwar et al 86 also showed that the 6 included RCTs all had a low risk of bias. Overall, it is suggested that future studies carefully consider the measurement of outcomes, register the RCT protocol, and report both significant and nonsignificant outcomes.

Our study has some limitations, which should be considered when interpreting the results. Although most of the participants were treatment-resistant patients with epilepsy, there was substantial heterogeneity in the study population in terms of age, severity of the disease, CBD dosage, source of CBD, and even its route of administration. The use of antiepileptic drugs, other than CBD, and the use of different dosages can influence the AEs that develop following the use of CBD. In addition, we only included AEs in the meta-analysis that were reported in at least 3 studies, so AEs that were only reported in 1 or 2 of the included studies were not reported. Furthermore, small-study bias and publication bias were not evaluated because fewer than 10 studies were included. 21 We could not investigate the association between CBD plasma levels and AEs, but this should be investigated in future RCTs and meta-analyses. In addition, we did not limit the selection criteria to only major RCTs, which might lead to the inclusion of studies in different phases, such as phase 2 studies with small numbers of patients and a short observation period. Furthermore, we did not perform a subgroup analysis by type of epilepsy, dose, or duration of the treatment because of the limited number of included studies and/or several studies not providing the relevant information.

In this systematic review and meta-analysis, the use of CBD to treat patients with epilepsy was associated with the development of several AEs, such as somnolence, diarrhea, decreased appetite, and AST or ALT elevation. Future research needs to investigate the therapeutic effects of CBD and AEs in the presence of various dosages of other antiepileptic drugs in order to achieve a safe and effective dose for treatment-resistant patients with epilepsy.

Accepted for Publication: March 7, 2023.

Published: April 20, 2023. doi:10.1001/jamanetworkopen.2023.9126

Corresponding Author: Saeid Safiri, PhD, Neurosciences Research Center, Aging Research Institute, Tabriz University of Medical Sciences, Golgasht Street, Azadi Street, Tabriz 5166614756, Iran ( [email protected] ); Koroush Gharagozli, MD, Brain Mapping Research Center, Loghman Hakim Hospital, Shahid Beheshti University of Medical Sciences, South Karegar Street, Tehran 1333635445, Iran ( [email protected] ).

Author Contributions: Drs Gharagozli and Safiri had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Drs Zahmatyar and ZareDini contributed equally to this work.

Concept and design: Gharagozli, Kolahi, Safiri.

Acquisition, analysis, or interpretation of data: Fazlollahi, Zahmatyar, Golabi, Sullman, Kolahi.

Drafting of the manuscript: Fazlollahi, Zahmatyar, ZareDini, Golabi, Nejadghaderi, Sullman, Gharagozli, Kolahi, Safiri.

Critical revision of the manuscript for important intellectual content: Fazlollahi, Nejadghaderi, Sullman, Kolahi, Safiri.

Statistical analysis: Fazlollahi, Zahmatyar.

Obtained funding: Gharagozli.

Administrative, technical, or material support: Nejadghaderi, Gharagozli, Kolahi, Safiri.

Supervision: Gharagozli, Kolahi, Safiri.

Conflict of Interest Disclosures: Dr Gharagozli reported receiving grants from the Brain Mapping Research Center during the conduct of the study. No other disclosures were reported.

Funding/Support: This study was supported by grant 43003046 from the Shahid Beheshti University of Medical Sciences, Tehran, Iran (Dr Gharagozli).

Role of the Funder/Sponsor: The funding source had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Data Sharing Statement: See Supplement 2 .

Register for email alerts with links to free full-text articles
Access PDFs of free articles
Manage your interests
Save searches and receive search alerts

Management of Psychogenic Nonepileptic Seizures

Information & authors, metrics & citations, view options, epidemiology, presentation and diagnosis.

Characteristic	Epileptic Seizures	Frontal Lobe Epilepsy	Psychogenic Nonepileptic Seizures

Writhing, flailing, and whole-body thrashing			Yes
Jactitation (rolling from side to side)		Yes	Yes
Lateral head and body turning		Yes	Yes
Eye-blinking, swallowing, and slumping			Yes

Intelligible speech			Yes
Eyes closed at seizure onset			Yes
Forced eye closure (resistance to the eyes being opened during an episode)			Yes
Postictal focal neurologic deficits	Yes
Altered breathing patterns	Yes
Somatic complaints	Yes
Increase in heart rate ≥30 bpm above baseline	Yes
Altered pupillary response	Yes

Conclusions

Key points/clinical pearls, information, published in.

Go to American Journal of Psychiatry Residents' Journal

Export Citations

If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download. For more information or tips please see 'Downloading to a citation manager' in the Help menu .

Format

Citation style
Style

To download the citation to this article, select your reference manager software.

There are no citations for this item

View options

Already a subscriber? Access your subscription through your login credentials or your institution for full access to this article.

Not a subscriber?

Subscribe Now / Learn More

PsychiatryOnline subscription options offer access to the DSM-5-TR ® library, books, journals, CME, and patient resources. This all-in-one virtual library provides psychiatrists and mental health professionals with key resources for diagnosis, treatment, research, and professional development.

Need more help? PsychiatryOnline Customer Service may be reached by emailing [email protected] or by calling 800-368-5777 (in the U.S.) or 703-907-7322 (outside the U.S.).

Share article link

Copying failed.

Request username.

Can't sign in? Forgot your username? Enter your email address below and we will send you your username

If the address matches an existing account you will receive an email with instructions to retrieve your username

Create a new account

Change password, password changed successfully.

Your password has been changed

Reset password

Can't sign in? Forgot your password?

Enter your email address below and we will send you the reset instructions

If the address matches an existing account you will receive an email with instructions to reset your password.

Your Phone has been verified

As described within the American Psychiatric Association (APA)'s Privacy Policy and Terms of Use , this website utilizes cookies, including for the purpose of offering an optimal online experience and services tailored to your preferences. Please read the entire Privacy Policy and Terms of Use. By closing this message, browsing this website, continuing the navigation, or otherwise continuing to use the APA's websites, you confirm that you understand and accept the terms of the Privacy Policy and Terms of Use, including the utilization of cookies.

Conversations with Leaders
Advanced search

Advanced Search

Psychogenic nonepileptic seizure: An empathetic, practical approach

Find this author on Google Scholar
Find this author on PubMed
Search for this author on this site
Find this author on Cleveland Clinic
For correspondence: [email protected]
Figures & Data
Info & Metrics

Psychogenic nonepileptic seizure (PNES) is often misdiagnosed as epilepsy, leading to unnecessary treatments and procedures, as well as failure to engage patients in needed mental health care. To establish an accurate diagnosis, video electroencephalography (EEG) in the context of and simultaneous with a comprehensive neurologic and psychosocial evaluation is recommended for any patient with seizures that are not responding to treatment. Delivering the diagnosis with empathy and respect is a crucial component of care that helps patients establish trust with caregivers and follow treatment recommendations. Effective treatment is available, highlighting the importance of early diagnosis to avoid unnecessary and potentially harmful treatment. But there are many barriers to care, including provider misperceptions, lack of acceptance of the diagnosis, poor patient engagement with treatment, and lack of access to care.

PNES resembles epileptic seizure in signs and symptoms but is due to psychological distress, a form of conversion disorder.

PNES is frequently misunderstood as being consciously feigned, and patients often feel accused of “faking” their seizures.

Inpatient video EEG in an epilepsy monitoring unit is the gold standard for diagnosis.

Psychotherapy should be tailored to the predisposing, perpetuating, and precipitating factors that contributed to the development of PNES.

A 19-year-old right-handed man who had meningitis at age 12 presented with seizures that had begun 12 months earlier. He described the seizures as bilateral arm-stiffening and stuttering speech, followed by rocking movements of the head and trunk that waxed and waned over 30 to 40 minutes. He said he never lost consciousness. He identified lack of sleep and stress as triggers.

The patient was brought to a local emergency department in the midst of a prolonged seizure and was treated with intravenous lorazepam. He was evaluated by a local neurologist, who prescribed levetiracetam for the seizures. Results of routine outpatient electroencephalography (EEG) and brain magnetic resonance imaging were normal. He continued to have seizures, despite escalation of levetiracetam doses.

See related article, page 260

He was admitted to the epilepsy monitoring unit for continuous video EEG monitoring. Several typical episodes were recorded and confirmed by family members and the patient. The episodes were characterized by gradual onset of irregular jerking of his head and arms, followed by arm and truncal stiffening and initial eyes-closed unresponsiveness. He then gradually started following commands but continued to have irregular bilateral jerking movements for 10 more minutes. No epileptiform EEG changes were seen before, during, or after the episodes. Likewise, interictal EEG over 72 hours was normal. He was diagnosed with psychogenic nonepileptic seizure (PNES).

PREVIOUSLY KNOWN AS PSEUDOSEIZURE

Previously known as pseudoseizure, PNES resembles epileptic seizures in symptoms and signs but is not caused by abnormal epileptiform electrical activity in the brain. Instead, this disorder is a manifestation of underlying psychological distress and unresolved emotions. Many people diagnosed with PNES meet the criteria for conversion disorder (also known as functional neurological symptom disorder) or other somatoform disorder, and others meet the criteria for dissociative disorder.

Multiple terms have been used to describe PNES, including dissociative seizure, functional seizure, stress seizure, and nonepileptic attack, reflecting the difficulty of finding a term that respectfully indicates both the psychological nature of the condition and its superficial similarity to epilepsy. The long-entrenched term pseudoseizure has been misinterpreted by patients and physicians as meaning the patient is “faking” or feigning the seizures. Unfortunately, this view has negatively influenced how some healthcare providers treat patients with PNES.

Importantly, there are other causes of nonepileptic events besides PNES—eg, syncope, migraine (which can be accompanied by transient focal neurologic symptoms and signs), paroxysmal dystonias, and other movement disorders. Rarely, a nonepileptic event is due to intentional deception as in factitious disorder or malingering. In some people with developmental or intellectual disabilities, nonepileptic events are behavioral or attention-seeking. PNES is distinctly different in that it is not conscious or intentional.

PATHOPHYSIOLOGY

The pathophysiology of PNES is unclear, but the literature suggests PNES is a network disorder affecting sensorimotor processing, emotional regulation, and neural responses to stress. 1 Functional neuroimaging studies provide some evidence that people with PNES have abnormalities in limbic brain structures including the amygdala, hippocampus, parahippocampal gyrus, insula, cingulate cortex, and prefrontal cortex. 2

EPIDEMIOLOGY

PNES can develop at any age but is most common between ages 15 and 35.

The disorder is more common in women, and particularly in women who have been victims of abuse. 3 Childhood abuse (sexual, emotional, or physical) is strongly correlated with subsequent development of PNES. 4 Psychiatric disorders such as depression, anxiety, and posttraumatic stress disorder (PTSD) are also commonly seen in patients with PNES, as discussed further below.

Early studies estimated the prevalence of PNES at 2 to 33 per 100,000. 5 A 2021 systematic review calculated the incidence of PNES in the United States at 3.1 per 100,000 per year, and the prevalence at 108.5 per 100,000. 6 In a 2021 population-based study in Norway, the mean annual incidence of PNES was also found to be 3.1 per 100,000 per year; the prevalence was 23.8 per 100,000, with the highest prevalence among 15- to 19-year-olds at 59.5 per 100,000. 7 In comparison, epilepsy has an incidence of 62 per 100,000 per year 8 and a prevalence of 1.2%, or 1,200 per 100,000. 9

From 25% to 35% of patients referred to epilepsy monitoring units for video EEG are diagnosed with PNES. 10 , 11 The disorder is often misdiagnosed as epilepsy, placing patients at risk of iatrogenic complications related to unnecessary antiseizure medications and inappropriate medical interventions such as intensive care unit admission, benzodiazepine administration, and oral intubation. In a study of 384 patients diagnosed with status epilepticus and treated unsuccessfully with benzodiazepines, 10% were ultimately determined to have PNES. 12

PNES is associated with poor quality of life 13 and high rates of unemployment and disability. 14 Mortality rates are also higher in people with PNES than in the general population, with one study finding that 20% of deaths in those with PNES under age 50 were due to suicide. 15

DIAGNOSED BY HISTORY AND VIDEO EEG

A comprehensive history and video EEG during a typical seizure are the gold standard for diagnosing PNES. There should be no epileptiform abnormalities on the EEG before, during, or after a typical event.

Absence of EEG changes alone, however, is not always diagnostic. EEG must be interpreted in the context of clinical signs and symptoms. Features of seizure semiology or symptomatology that are highly predictive of PNES include long duration of convulsive-type seizures (> 10 minutes), convulsive-type seizures with retained awareness, rapid side-to-side head movements, out-of-phase limb movements, eyes-closed unresponsiveness, and pelvic thrusting ( Table 1 ). 16 Fluctuating patterns of movement and distractibility during the seizure are also suggestive of PNES.

View inline

Clinical features that may suggest psychogenic nonepileptic seizure a

No one sign is 100% specific for PNES . For instance, out-of-phase limb movements and pelvic thrusting can occur in frontal lobe epileptic seizures, without a clear ictal EEG change.

Video EEG is most helpful when there are motor signs or decreased responsiveness, but like most diagnostic tools, video EEG has limitations. For instance, if the onset of the seizure is not captured on video, postictal behavior can be confused with PNES.

Importantly, video EEG is less useful when the patient has only subjective symptoms, because epileptic aura (with purely subjective symptoms) can be scalp EEG-negative. In addition, certain epileptic seizures can be scalp EEG-negative due to movement artifact or because scalp EEG has difficulty recording from deeper areas of the brain. In these cases, referral to a comprehensive epilepsy center is recommended. As mentioned earlier, other nonepileptic events to consider are migraine, vertigo, syncope, movement disorder (eg, paroxysmal dystonia and dyskinesia), and sleep disorders such as narcolepsy, cataplexy, and parasomnias.

About 10% of patients with PNES also have epileptic seizures, so when the patient or the patient’s family describes more than 1 seizure type, it is crucial to record examples of all seizure types. Once it is confirmed that a patient has both PNES and epileptic seizures, showing the patient and family videos of the seizure types captured with video EEG, and highlighting key features of both seizure types, will help them distinguish PNES from epileptic seizures once they leave the monitoring unit.

COMMUNICATE THE DIAGNOSIS CLEARLY AND WITH EMPATHY

Presenting the diagnosis to the patient is typically the job of the neurologist who has interpreted the video EEG. Communicating the diagnosis effectively is crucial and can be therapeutic in the short term. However, if learning the diagnosis leaves the patient angry or confused, PNES and other psychiatric symptoms will likely worsen.

A survey of primary care and emergency medicine physicians found that 38% believed that episodes of PNES are intentionally produced or faked, and 63% did not feel video EEG was needed to confirm a diagnosis of PNES. 17 The misperception that PNES is intentionally feigned is likely to result in mismanagement of the condition.

Many patients with PNES say the diagnosis is confusing and distressing, and they feel misunderstood, mistreated, and blamed when they seek medical care. 18 About a quarter feel the diagnosing doctor does not understand their PNES symptoms. 19 Receiving a diagnosis of PNES can be particularly confusing for patients who were previously diagnosed with epilepsy and treated for years with antiseizure medications. 20 When their diagnosis is changed from epilepsy to PNES, patients find the news distressing because they perceive the burden of recovery is shifted from the doctor’s shoulders to theirs. 21 Misperceptions about PNES and poor physician-patient communication certainly add to the emotional struggles of patients and can lead to resistance to mental health recommendations.

Since many people with PNES have a history of trauma and abuse, perceived or actual mistreatment by medical providers (via poor communication of the diagnosis) can traumatize them yet again and makes it more likely they will reject the diagnosis. Various communication strategies have been proposed, but the most important component is to deliver the diagnosis with empathy and clarity.

Key points in discussing the diagnosis with the patient are to acknowledge that their symptoms are real and can be frightening and disabling. It can be reassuring to know that they are not alone and that PNES is a diagnosis that is common in epilepsy monitoring units.

The discussion should also clarify that the patient does not have epilepsy and does not need antiseizure medications (assuming the patient does not have comorbid epileptic seizures). Rapid titration off antiseizure medications at the time of diagnosis is associated with better outcome than with delayed titration. 22

It is helpful to discuss the role of emotions and stress in producing physical symptoms, similar to the way anxiety can cause abdominal pain or headaches. Finally, it is essential to let the patient know that with treatment PNES can resolve, and that seizure control with a return to normal function should be the goal. These steps are summarized in Figure 1 .

Download figure
Open in new tab
Download powerpoint

Algorithm for diagnosing psychogenic nonepileptic seizure (PNES).

Emergency management

The basics of emergency medical care apply in people having a known or suspected PNES episode, as follows:

Monitor airways, breathing, and circulation

Provide for patient safety and comfort

Avoid employing noxious stimuli (eg, sternal rub) in an attempt to test responsiveness

Remain calm and reassuring

Stay with the patient until symptoms start to improve.

If the PNES diagnosis is clear from a previous video EEG evaluation and if the situation allows, encouraging the patient to engage in deep breathing can help to lessen the intensity of the episode. Once the episode has resolved, prompting the patient to identify potential triggers for the episode can be instructive and ultimately empowering.

If the seizure diagnosis is not clear, PNES should still be considered, if only briefly, before initiating escalating doses of antiseizure medications in an emergency setting.

Predisposing, precipitating, and perpetuating factors

Biologic, psychological, and social factors all contribute in a complex way to predisposing patients to PNES, precipitating episodes, and perpetuating the condition, thus making it chronic ( Figure 2 ).

A variety of predisposing, precipitating, and perpetuating factors contribute to psychogenic nonepileptic seizure (PNES). Patients with PNES typically have multiple contributing factors.

Biologic factors include a history of head injury and of somatic conditions such as migraine, asthma, irritable bowel syndrome, chronic pain, and insomnia.

Psychological factors associated with PNES include mood disorder, anxiety, PTSD, and maladaptive coping styles. Exposure to trauma early in life can contribute to the emergence of psychiatric symptoms such as somatic dissociation due to inability to regulate emotions and cope with distress. Maladaptive coping styles, particularly the avoidant coping style and alexithymia (inability to identify and describe emotions), can make people susceptible to develop somatic symptoms as a means to release tension. Heightened somatic hypervigilance, excessive symptom preoccupation, and learned somatization can all contribute to the development of PNES. 23 , 24

Social factors include a history of abuse, chronic stress, drug use, family dysfunction, marital discord, and financial instability.

A single factor can play multiple roles, both predisposing to and perpetuating PNES. Typically, a combination of biopsychosocial factors including physiological susceptibility, early-life trauma, maladaptive response to psychological distress, and ongoing social stressors can lead to the development and chronicity of PNES. 25

Psychiatric disorders: Cause or comorbidity?

Symptoms of PNES are considered maladaptive defense mechanisms that develop in response to an underlying psychiatric disorder. 26 Therefore, coexistent psychiatric disorders can be understood as causes of PNES rather than comorbidities. This relationship can be bidirectional, with psychiatric symptoms contributing to the emergence of PNES, and the struggle with PNES exacerbating existing psychiatric disorders. Therefore, the assessment and treatment of PNES should include identifying and addressing coexisting psychiatric disorders along with the PNES symptoms.

Common psychiatric comorbidities in patients with PNES include the following 27 :

PTSD (35%–49%)

Depressive disorders (57%–85%)

Dissociation (22%–91%)

Other somatoform disorders (22%–84%)

Axis II (personality) disorders (10%–86%).

Suicidal ideation is common in individuals with PNES, with 39% acknowledging suicidal ideation and 20% reporting suicide attempts in 1 study. 28 Panic attacks, history of trauma, and history of sexual and physical abuse are also highly prevalent.

The high prevalence of trauma exposure and psychiatric comorbidity reflects the extreme vulnerability and psychological distress that patients with PNES suffer and helps explain the critical need for psychological support. A misperception of the condition as consciously feigned slights the patient’s struggle, increases distress, and worsens PNES symptoms.

Psychotherapy is effective

PNES is treatable, as demonstrated by 2 pilot randomized controlled trials of 12-session courses of cognitive behavioral therapy (CBT). 29 , 30 A meta-analysis of psychological interventions including CBT found that 47% of patients with PNES became seizure-free, and 82% showed a reduction in seizures of at least 50%. 31 PNES-tailored counseling interventions, particularly CBT-based, also improve health-related quality of life and psychosocial functioning. 32

PNES-specific counseling interventions often include education about types of seizures, identifying and managing common seizure triggers, aura interruption methods, and improved emotion management skills using relaxation training and other CBT techniques. 29

As mentioned earlier, controlling underlying psychiatric symptoms is an important part of treating PNES. In the case of ongoing psychiatric symptoms such as PTSD, various evidence-based psychotherapy interventions can be used concurrently or subsequently, including the following:

Eye movement desensitization and reprocessing

Prolonged exposure therapy for patients with coexisting PTSD symptoms 33

Cognitive processing therapy

Intensive outpatient programs for mood disorders

Dialectical behavioral therapy for patients with severe personality disorders 34

Family therapy, often incorporated in individual counseling because of the high prevalence of family dynamic stress in patients with PNES. 35

Pharmacologic therapy for some

Although counseling is the best intervention, antidepressants are often used to treat PNES, particularly in patients with low psychological insight or poor engagement with counseling for other reasons. 29 – 32 There is some evidence that antidepressants alone, 36 as well as antidepressants with counseling, 29 can result in reduction of PNES episodes.

The benefit from benzodiazepines is mixed. Although some patients may benefit from benzodiazepines for anxiety, clobazam and clonazepam have been associated with behavioral side effects that can mimic PNES. 37

Stop antiseizure medications

Continuing antiseizure medications in patients with PNES has been associated with poor outcome. 38 When the diagnosis of PNES is clear, antiseizure medications should be stopped unless they are being used to manage comorbid epilepsy, chronic pain, migraine, or mood instability.

IMPROVING TREATMENT ADHERENCE

Effective treatments are available for PNES, but challenges remain, especially lack of access to treatment and patient rejection of both the diagnosis and treatment.

High attrition and poor treatment engagement are known challenges in the treatment of PNES. Predictors of poor treatment adherence include insufficient understanding of the diagnosis, unemployment, and severe psychiatric and personality disorders. 28 , 39 Communicating the diagnosis without sufficient explanation or a clear treatment path rarely produces a good outcome, whereas patients who are given sufficient time and education about the diagnosis, as well as psychiatric support, show better outcomes. 40 , 41

Although physicians have no control over the patient factors that predict poor treatment engagement, they do have control over how they explain the diagnosis, which in turn can affect the patient’s acceptance of the diagnosis, which is the first step in treatment engagement. Introducing the diagnosis may initially invoke intense emotions in patients, but taking sufficient time to explain PNES and answer questions, using an empathic approach to validate patients’ reactions, can help ease patient distress. Recognizing that shame and embarrassment are common reactions in these situations, a dignified and respectful conversation during the delivery of the PNES diagnosis can help the patient to be receptive of the physician’s recommendations.

The psychological approach known as motivational interviewing, often used to engage treatment-resistant patients, was shown in a randomized control trial to improve patients’ acceptance of the diagnosis, adherence to treatment, and quality of life, as well as to reduce the frequency of PNES episodes. 42 Empathic and clear communication of the diagnosis and allowing sufficient time to address all of the patient’s concerns and questions are critical components of the treatment of PNES.

MORE ABOUT OUR PATIENT

We talked to the patient further and found that he began to have depressive symptoms after his grandmother died, 4 years before the onset of his seizures. In the year after her death, he began to drink alcohol and abuse drugs.

After graduating from high school in May 2020, he joined the military, but soon after, he tested positive for COVID-19 and was placed in quarantine. Being diagnosed with COVID-19 early in the pandemic when there was so little information available was a traumatic experience for him. He felt helpless and had severe crying spells because he thought he was going to die. His quarantine “buddies” were likewise experiencing depressive symptoms, and he witnessed multiple episodes of self-injurious behavior among the other recruits. While in quarantine, he developed seizures and was hospitalized.

He was eventually discharged from the military and returned home. He then enrolled in college, where he struggled with his classes and had a series of failed romantic relationships.

In the epilepsy monitoring unit, he was diagnosed with anxiety in addition to PNES. The diagnosis of PNES was explained in the context of his recent stressors, and though he was tearful, he said he felt relieved to know he did not have epilepsy. He and his family understood and accepted the PNES diagnosis, and outpatient psychotherapy was scheduled.

DISCLOSURES

The authors report no relevant financial relationships which, in the context of their contributions, could be perceived as a potential conflict of interest.

Copyright © 2022 The Cleveland Clinic Foundation. All Rights Reserved.
Balachandran N ,
Goodman AM ,
Allendorfer JB , et al
Szaflarski JP ,
LaFrance WC Jr .
Asadi-Pooya AA ,
Valente K , et al
Benbadis SR ,
Allen Hauser W
Asadi-Pooya AA
Villagrán A ,
Aaberg KM ,
Maloney EM ,
O’Reilly ÉJ ,
Costello DJ
Salinsky M ,
Spencer D ,
Boudreau E ,
Chen-Block S ,
Abou-Khalil BW ,
Arain A , et al
Chamberlain JM , et al
Goldstein LH ,
Robinson EJ ,
Reuber M , et al
Nightscales R ,
McCartney L ,
Auvrez C , et al
LaFrance WC Jr . ,
Shneker BF ,
Rawlings GH ,
Chaudhary F , et al
Fernández G ,
Helmstaedter C ,
Karterud HN ,
Knizek BL ,
Witthöft M ,
Elliott JO ,
Negrini PB ,
Perin C , et al
Ettinger AB ,
Devinsky O ,
Weisbrot DM ,
Ramakrishna RK ,
Barry JJ , et al
Chalder T ,
Chigwedere C , et al
Carlson P ,
Nicholson Perry K
Mellers JDC , et al
Vaidya-Mathur U ,
Linehan MM ,
Comtois KA ,
Murray AM , et al
Krawetz P ,
Fleisher W ,
Keitner GI ,
Papandonatos GD , et al
Massot-Tarrús A ,
AlKhateeb M ,
Mirsattari SM
Hall-Patch L ,
House A , et al
LaRoche SM ,
Ganesh GA ,
Teagarden D ,
Tolchin B ,
Martino S , et al

In this issue

Cleveland Clinic Journal of Medicine: 89 (5)

Table of Contents
Table of Contents (PDF)
Index by author
Complete Issue (PDF)

Thank you for your interest in spreading the word on Cleveland Clinic Journal of Medicine.

NOTE: We only request your email address so that the person you are recommending the page to knows that you wanted them to see it, and that it is not junk mail. We do not capture any email address.

Citation Manager Formats

EndNote (tagged)
EndNote 8 (xml)
RefWorks Tagged
Ref Manager

Tweet Widget
Facebook Like
Linkedin Share Button

Jump to section

Cited By...

Psychogenic nonepileptic seizure: A neurologists perspective

Recent advances in epilepsy

Mark manford.

Department of Clinical Neurosciences, Addenbrooke’s Hospital and University of Cambridge, Hills Rd, Cambridge, CB2 0QQ UK

This paper reviews advances in epilepsy in recent years with an emphasis on therapeutics and underlying mechanisms, including status epilepticus, drug and surgical treatments. Lessons from rarer epilepsies regarding the relationship between epilepsy type, mechanisms and choice of antiepileptic drugs (AED) are explored and data regarding AED use in pregnancy are reviewed. Concepts evolving towards a move from treating seizures to treating epilepsy are discussed, both in terms of the mechanisms of epileptogenesis, and in terms of epilepsy’s broader comorbidity, especially depression.

Definitions and classification

Definitions in epilepsy have always been problematic [ 1 – 5 ]. The disorder is characterised by seizures but not all seizures are due to epilepsy—febrile seizures or drug induced seizures, for example. Earlier classifications sought to reconcile these difficulties by describing different electroclinical syndromes but new data from modern imaging and genetics need to be incorporated.

Diagnosis is difficult because in practice, the diagnostic electrical hallmark of epilepsy may be absent interictally, especially in adults or if seizures are infrequent and interictal epileptiform discharges may occasionally be present in those without seizures. Moreover, in some instances, an “epileptic EEG” may be associated with an epileptic encephalopathy, in which overt seizures may be few or none, such as Landau–Kleffner syndrome, and a cognitive disorder dominates the presentation.

The International League Against Epilepsy recently consulted in an attempt to synthesise a consensus view [ 6 ], whose output will be published in 2017. The result promises to be useful and pragmatic, recognizing that the syndromes are multifaceted; any one case defined by an association of clinical, electrophysiological, etiological and comorbid factors. It also accepts that it is not always known if seizures are part of focal or generalized epilepsy and that in some cases, such as tuberous sclerosis, genetic and structural causes overlap. Some terms will be dropped, for example, childhood epilepsies where the seizures remit will be called pharmacoresponsive rather than benign, recognizing that children whose seizures remit may nevertheless have significant persisting psychosocial comorbidities.

The ILAE has also pondered the question of whether a single seizure may be considered to be epilepsy [ 7 ] and concluded that it may if there is a greater than 60% chance of another seizure; a risk conferred by the presence of EEG spikes or a major structural aetiology. Epilepsy may be considered to have gone away after ten years with no seizures and with no treatment. This approach has pragmatic utility, rather than mechanistic validity and is useful in allowing driving regulatory authorities to treat those with lower risk more leniently and may be helpful in deciding when to treat medically after a single seizure [ 8 ].

Some frontal lobe epilepsies may be particularly difficult to diagnose, often with non-diagnostic ictal scalp EEGs and some were initially considered to be a movement disorder, e.g. “paroxysmal nocturnal dystonia” [ 9 ] in which its epileptic basis was shown later [ 9 , 10 ]. The situation has become more complex with the discovery that patients with frontal lobe epilepsy may also have epileptic nocturnal wandering, with similarities to parasomnias and also brief nocturnal movements which are not due to seizure discharges but may be a release phenomenon of interictal discharges [ 11 ]. They may suffer also from non-epileptic parasomnias more frequently than the general population. In the new classification, the phenomenon will be renamed “Sleep-related hypermotor epilepsy (SHE)”.

Status epilepticus and limbic encephalitis

The ILAE recently defined status epilepticus as: “a condition resulting either from the failure of the mechanisms responsible for seizure termination or from the initiation of mechanisms, which lead to abnormally, prolonged seizures (after time point t 1 ). It is a condition, which can have long-term consequences (after time point t 2 ), including neuronal death, neuronal injury, and alteration of neuronal networks, depending on the type and duration of seizures” [ 12 ]. Timepoint t 1 is at 5 min after seizure onset, when it is recognized for generalized tonic–clonic status epilepticus that evolution to status is increasingly likely and when treatment should be initiated. T2 is at 30 min, after which there is increasing risk of irreversible consequences. Status is divided along four axes; semiology, aetiology, EEG correlates and age. These axes align with the prognosis of status, which when adequately treated is determined by cause and the age and gender of the patient. The electroclinical state is another prognosticator; subtle status evolving from convulsive status has a particularly poor prognosis [ 13 , 14 ].

The impressive out-of-hospital randomized, double-blind RAMPART study has shown that IM midazolam is at least as effective as IV lorazepam in the early treatment of status, in adults and children [ 15 , 16 ], probably because IM speed of administration of midazolam compensates for speed of IV distribution of lorazepam. It has long been known that the effect of benzodiazepines in status epilepticus wears off very rapidly [ 17 , 18 ] and it has subsequently been demonstrated that GABA A receptor sensitivity is reduced, sometimes long term [ 18 ]. Receptor trafficking may be contributory [ 19 , 20 ]. As well as a reduction in inhibitory neurotransmitters, within 1 h of onset of status in rats, there is an increase in surface NMDA receptors in status, associated with increased excitation [ 21 ]. Cholinergic mechanisms are also implicated, supported by the observation that in pilocarpine induced status epilepticus; the addition of scopolamine provides additional seizure control, when combined with phenobarbital and benzodiazepines, raising the possibility of the use of drug combinations in status [ 22 ].

Basic mechanisms are starting to align with clinical evidence in the initial treatment of status with benzodiazepines, but thereafter the evidence is less clear. Initial uncontrolled reports suggested a 70% success rate for the treatment of status epilepticus with levetiracetam [ 23 ], but a recent randomized controlled trial of out-of-hospital clonazepam plus either levetiracetam or placebo was abandoned because of a lack of benefit in the levetiracetam arm [ 24 ]. This mirrors the finding that diazepam plus phenytoin confers no additional benefit to lorazepam alone at 12 h [ 14 ] and raises questions around the appropriate timing of the addition of AED to benzodiazepines. It also emphasizes the importance of properly controlled studies in an area where few have been undertaken. Shorvon et al. have undertaken meta-analyses of existing therapies [ 25 – 27 ]. From generally poor quality studies of lacosamide, levetiracetam, phenobarbital, phenytoin or valproate in benzodiazepine resistant status, they found efficacy ranging from 50% (phenytoin), to levetiracetam (68.5%), phenobarbital (58–84%) and valproate 76%. Lacosamide treatments were too few to give figures. The conclusion remains that all these drugs may be useful but there is no clear guidance on choice. The caution with which data from uncontrolled studies must be interpreted is highlighted by a recent randomized study of valproate versus phenobarbital which showed a 44% response to valproate and an 81% response to phenobarbital. However, in children, valproate may have fewer adverse effects and better efficacy than phenobarbital [ 28 , 29 ] and similar efficacy to phenytoin [ 30 ]. But children may not be comparable to adults with a greater proportion of generalized epilepsies, more responsive to valproate. Future options include derivatives of valproate such as valnactomide and butylpropylacetamide, which may be more potently antiepileptic and less teratogenic in animal studies [ 31 ].

For status epilepticus which remains refractory to a second line AED, a range of intravenous benzodiazepines or anaesthetic agents may be considered and again Shorvon et al. found that studies are of poor quality. They found that 35% of patients in these studies died and a further 13% had severe neurological deficits and 13% mild neurological deficits on recovery. Studies underway may help answer some of these questions [ 32 , 33 ]. Ketamine’s role in blocking NMDA receptors [ 34 ] has led to it become increasingly popular in the treatment of refractory status, with some efficacy on the basis of uncontrolled retrospective series [ 35 – 37 ]. A randomized trial in children is planned [ 38 ]. A recent trial of hypothermia showed no benefit at 90 days [ 39 ].

It is increasingly recognized that some patients with refractory status epilepticus, where the cause was previously unrecognized, may be suffering from an antibody mediated encephalopathy, “limbic encephalitis”. Antibodies implicated include LGI1 and NMDA, with CASPR less associated with seizures [ 40 , 41 ]. More recently, GABA B and AMPA receptors have been implicated in some cases [ 42 ]. A specific phenotype of very brief, frequent and highly focal, faciobrachial dystonic seizures is almost pathognomonic of LGI1 associated disease, often heralding a more severe encephalopathy [ 43 ] and providing an opportunity to intervene at an earlier stage. Limbic encephalitis exhibits characteristic changes on MRI in the mesial temporal structures, especially the amygdalae [ 44 ] and responds primarily to immunotherapy and treatment of any associated tumour, rather than to AED [ 41 , 45 ]. Early suspicion of the diagnosis and treatment, even before definitive serological confirmation, is recommended. Many patients will recover with appropriate treatment but may be left with ongoing epilepsy and hippocampal sclerosis is a reported outcome [ 46 ]. The extent to which epilepsy in patients, who have not suffered limbic encephalitis, may be attributable to antibody-mediated disease is an area of exploration which may open new avenues of treatment for chronic epilepsy. Small cohorts suggest increased rates of antibody positivity but their significance is not yet clear [ 47 , 48 ].

Pharmacological treatment of epilepsy and underlying mechanisms/genetics

In 2000, Kwan and Brodie [ 49 ] found that 63% of unselected patients in an epilepsy service were rendered seizure free with medication. Since then despite numerous antiepileptic drugs becoming available, they found that the chance of a patient, who is diagnosed in 2017 becoming seizure free, has changed little [ 50 ]. Some studies are more optimistic; refractory epilepsy may have a greater chance of 12-month remission with or without AED change [ 51 – 53 ] at around 5% per year and although up to 40% may relapse [ 51 ], many of these may have a second longer remission.

The broad sweep of AEDs, generally affecting ion channels or neurotransmitters is unchanged, but there is slowly increasing evidence for a differential effect in specific syndromes.

Of established epilepsy drugs, ethosuximide, often forgotten by adult neurologists, has the most specific mechanism in relation to its role in the absence epilepsy. It acts on T-type calcium channels [ 54 ], implicated in the thalamocortical disturbance believed for decades to underlie generalized epilepsies [ 55 ]. Valproate and ethosuximide have clearly demonstrated greater efficacy over lamotrigine in childhood absence epilepsy [ 56 ]. A small, non-randomized study has suggested that ethosuximide may be also associated with a greater chance of long-term remission [ 57 ]. In a mouse model of absence epilepsy, Bomben et al. [ 58 ] selectively ablated P/Q channels in the neurons of layer VI that provide the descending cortical projection to the thalamus. This produced spike-wave activity with clinical absences suppressed by ethosuximide. This very selective lesion supports the view that a highly specific cortical abnormality is necessary and sufficient to generate the thalamocortical oscillations of absence epilepsy. Not all patients respond equally to medication. A clinical imaging and EEG study, comparing those patients responsive to valproate to those who are resistant, suggested different patterns of activation may underlie the varying therapeutic responses [ 59 ].

Despite strong epidemiological evidence of a genetic basis of IGE, relevant genes remain elusive, hampering efforts to identify specific drug targets. A recent genome wide association study suggested links to SCN1A, a known cause of GEFS+, protocadherin PCDH7 and PCDH19, both known to be associated with epilepsy and learning disability [ 60 ]. An analysis of microdeletions in generalized epilepsy showed an increased burden (7.3%) compared to controls (4%) and specific involvement of a range of genes known to be important in epilepsy, psychiatry and neurodevelopment [ 61 ].

The first major application of pharmacogenetics in epilepsy, and probably still the most widely applicable, has been the identification of patients from South East Asia who are HLA-B*1502 positive, putting them at high risk for Stevens–Johnson syndrome from carbamazepine and the elimination of this life-threatening complication by pre-treatment screening [ 62 , 63 ]. Genetic understanding is creeping into other areas of pharmacological therapeutics. It has been realized for a number of years that sodium channel blocking drugs may be deleterious for children with Dravet syndrome [ 64 , 65 ], although this may not be so clear for adult patients [ 66 ]. It is now known that Dravet syndrome is commonly due to a genetic truncations leading to total loss of function or missense mutations causing partial loss of function of the sodium channel, usually SCN1A [ 67 , 68 ], which is located on inhibitory interneurons and causes hyperexcitability and seizures as a result of loss of function. A previously empirical observation of relative AED efficacy is now underpinned by a mechanistic understanding, which can guide drug choice. Mutations of the SCN8A gene are also associated with epilepsy, sometimes with a Dravet-like syndrome [ 69 ]. However, the phenotype may depend on the pathophysiology of the mutation, which may be a gain or a loss of function [ 70 ]. In four children with epileptic encephalopathy onset in the first months of life, Boerma described a response to phenytoin [ 71 ]. One of these had been demonstrated to have a gain of function mutation.

There are a number of other instances where rare monogenic cases of epilepsy have been evaluated in detail and treatment tailored to the identified pathophysiological mechanism, with varying success. Most consistently effective is the use of ketogenic diet to switch cerebral energy metabolism away from glucose in patients with Glut-1 deficiency, which may be dramatically successful [ 72 , 73 ]. Retigabine (ezogabine) increases activity at KCNQ2 channels [ 74 ] and has been used to treat the neonatal epileptic encephalopathy associated with reduced function mutations of the KCNQ2 channel with some success [ 75 ]. Unfortunately, this drug is to be withdrawn from use in 2017 because of the pigmentary changes it may induce in skin, mucosae and eyes [ 76 ]. GRIND2 mutations resulting in gain of activity of the NMDA receptor may cause balloon swelling and cell death. Children with a severe encephalopathy due to this mutation may possibly benefit from treatment with memantine, more generally used in Alzheimer’s disease which inhibits this channel [ 77 ]. KCNT1 encodes a sodium-activated potassium channel and has been implicated in the migrating partial epilepsy of childhood and in autosomal dominant frontal lobe epilepsy, both causing a gain of function [ 78 ]. Two children with this mutation and a severe epilepsy phenotype were helped by the administration of quinidine [ 79 ]. These cases illustrate the importance of not only an electroclinical and genetic diagnosis of these epilepsies but also delineation of the specific pathophysiology of the mutation to enable drug choice, which may include opportunities beyond those conventionally used in the antiepileptic armamentarium.

Epileptogenesis and inflammation

Another focus is the mechanisms of epileptogenesis; the process from initiation of pathological changes to the development of epilepsy and possibly the maintenance of epilepsy. There are changes, which involve altered gene expression, inflammation, protein production and changes in connectivity, which may all be the target for drugs to suppress epileptogenesis. One of the most studied pathways links to the rapamycin (mTOR) pathway (Fig. 1 ). Upregulation of mTOR, a serine/threonine protein kinase, occurs as a result of the TSC1 and TSC2 mutations of tuberous sclerosis (TS) complex. Other mutations in the pathway may be associated with overgrowth in megalencephaly [ 80 ]. mTOR has a role in protein synthesis and inhibition of mTOR, cell growth and replication by everolimus, a rapamycin analogue, has been shown to reduce overgrowth of malignantly transformed tubers [ 81 ]. Animal models have shown an antiepileptic effect of mTOR inhibition [ 82 ] but this has been more difficult to demonstrate in humans. However, a recent double-blind study of 366 patients showed a dose-related seizure reduction of up to 40% with everolimus, in patients with TS [ 83 ]. However, mTOR inhibitors may also have a direct effect on Kv1.1 ion channels, independent of epileptogenesis [ 84 ], blurring their possible mechanism in seizure suppression.

An external file that holds a picture, illustration, etc.
Object name is 415_2017_8394_Fig1_HTML.jpg

Pathway showing some of the relationships between mTOR and cellular function which may be modulated in epileptogenesis and their modulation through inflammatory pathways and by drugs. AMPK 5′ AMP-activated protein kinase, IRS1 insulin receptor substrate 1, JAK Janus kinase, MTOR mechanistic target of rapamycin, PDK1 pyruvate dehydrogenase lipoamide kinase isozyme 1, P13K PI3 kinase, PKB protein kinase B, PtdIns phosphatidylinositol, PTEN phosphatase and tensin homologue, RHEB ras homolog enriched in brain (GTP binding protein), STRADA STE20-related kinase adaptor alpha, STK11 serine/threonine kinase 11, TSC tuberous sclerosis complex

Whilst immunological mechanisms are clearly implicated in the aetiology of certain epilepsies such as limbic encephalitis [ 85 ] or Rasmussen encephalitis [ 86 ], increasing attention has been given to them in commoner forms of epilepsy. There is broad evidence for their significance, especially from animal studies and involving cytokines, changes in the blood brain barrier and pathological alterations associated with altered excitability [ 87 – 94 ]. Pathological examination of resected human specimens of focal cortical dysplasia [ 95 ] has also shown substantial increases in mRNA expression of Toll-like receptors 2 and 4 and associated with high-mobility group box protein 1, restricted to astrocytes and microglia in pathological tissue. These interact through interleukin IL1-β. Microglia activation appears increased more in focal cortical dysplasia (FCD) type II than in FCD I, associated with the migration of activated lymphocytes and activation of the mTOR pathway, linking inflammation to epileptogenesis [ 96 ]. A recent systematic review and meta-analysis [ 97 ] has described increased CNS levels of interleukins of the IL1 family as well as of chemocytokines CCL 3-5, which are involved in monocyte and lymphocyte migration. IL6 appears to be elevated in serum but not in CNS. A recent study of patients with moderate to severe cerebral trauma found a relationship between cerebrospinal fluid IL1-β levels and an allelic variant of the IL1-β gene to the risk of developing epilepsy [ 98 ]. This provides the first evidence of a biomarker that might be used to predict epilepsy after an epileptogenic insult and possibly a means of pharmacological intervention. These may need to be complex; a recent study suggested a single intervention was inadequate and a cocktail of anti-inflammatory drugs was required to prevent epileptogenesis [ 99 ]. A small case series of intractable childhood onset epilepsy has already been treated successfully with human recombinant IL-1 receptor antagonist (Anakinra\) [ 100 ] and it is hopeful that, as there are already many drugs affecting the immune system and some affecting the blood brain barrier, that this will prove a fertile area for development.

Recently, mutations of the DEPDC5 (DEP domain containing 5, involved in g-protein signalling) gene have been demonstrated in patients with cortical dysplasia and in up to 12% of small families of patients with familial focal epilepsy phenotypes, including ADNFLE without demonstrable lesions [ 101 – 103 ]. This gene is involved in the same GATOR pathway as mTOR. Although the GATOR (gap activity towards RAG’s) pathway is generally associated with protein synthesis, it appears to reduce the levels of K v 1.1 potassium channels in hippocampal pyramidal neurons increasing seizure expression, which can be reversed by inhibitors [ 104 ]. These findings link lesional and non-lesional ion channel related epilepsies to the same pathway, providing a potential opportunity for the wider use of inhibitors in treatment.

Although the scope is expanding, the relationship of these mechanisms to the majority of epilepsies, those triggered by a neurological insult (focal epilepsies) or a complex genetic trait (generalized epilepsies) remains to be established. It has long been recognized that epilepsy due to trauma is more likely in those with a family history of epilepsy [ 105 ] providing a potential to link to genetic mechanisms. But the development of epilepsy may take 20 years [ 105 , 106 ]. The key will be to identify those patients at high risk and to find a low risk preventative treatment akin to aspirin in stroke and very large, long-term follow up studies, will be needed to establish efficacy. Biomarkers such as IL1-β for evolving epileptogenesis are needed to identify high risk patients and to act as drug targets.

Antiepileptic drug trials

Despite being a common disorder, the number of high quality trials of antiepileptic drugs is small. Trials of new AED are normally in the form of an add-on therapy in refractory partial epilepsy, usually with the end point of a 50% reduction of seizures. This may be realistic in showing a biological effect but does not confer the psychosocial benefits of seizure freedom, and therefore drugs enter the market with the knowledge that they will not dramatically alter the burden of refractory epilepsy. The Federal Drug Administration in the US requires monotherapy trials against placebo and the European Medicines Agency requires head-to-head trial of active agents. Consequently, results cannot cross the Atlantic, delaying introduction and increasing cost for manufacturers. Both types of trials have their merits. The result is a non-systematic hotchpotch of evidence in relation to monotherapy in epilepsy. Whilst the pragmatic study SANAD has guided many UK clinicians to lamotrigine as first line in monotherapy for focal epilepsy [ 107 ], carbamazepine remains a drug of choice in many countries and studies [ 108 ]. A recent study has shown that zonisamide is non-inferior to carbamazepine in new onset focal epilepsy in adults [ 109 ]. A large study of 1688 new onset patients compared time to withdrawal of levetiracetam in two arms to first choice carbamazepine or valproate in monotherapy in adults [ 110 ]. Overall, the drugs performed similarly but in a post hoc analysis, levetiracetam withdrawal rate was lower in those over 60, especially in comparison to carbamazepine, with fewer adverse effects rather than greater efficacy [ 111 ].

The repertoire of AED considered effective in IGE has traditionally been more restricted that for focal epilepsy. Case reports have supported the use of lacosamide [ 112 , 113 ] and it is the subject of ongoing larger scale studies. Perampanel has been found to be effective as an add-on for refractory generalized epilepsy with tonic–clonic seizures [ 114 ].

Cannabis contains approximately 80 different active cannabinoids and was used in the nineteenth century as an AED [ 115 ]. It has been known for many years to be an antagonist at NMDA receptors with antiepileptic activity [ 116 ]. Δ 9 tetrahydrocannabinol is the main psychoactive component of cannabis, acting on THC1 and THC2 receptors but other components, especially cannabidiol (CBD) do not act on these receptors, are not psychoactive. They may have medicinal properties through a range of other actions [ 117 ]. Clinical studies in the 1970s and 80s reviewed in [ 117 ] pointed to antiepileptic effects and recent anecdotal evidence and an open labelled trial have shown benefit in epileptic encephalopathies such as Dravet syndrome [ 118 , 119 ], which have had a profound social effect in the United States, with parents moving their families to states where cannabis is legal [ 120 ]. Although their mechanisms point to a potential role for cannabinoids of relevance to epilepsy [ 121 ], there are as yet, no good studies to support their widespread use. The adverse effects of natural cannabis are widely known [ 122 ] and a particular problem for adolescents. Cannabinoids should be avoided by those with epilepsy, especially the young, who are already at risk of psychiatric problems, until good quality trials support their use.

Epilepsy and comorbid depression

Data extracted from a US population survey of 340,000 households and those with epilepsy were compared to those without [ 123 ]. Two percent had suffered with epilepsy and reported increases in a range of disorders (Table 1 ). A figure of approximately one third affected by depression is consistent with numerous previous studies. The relationship to epilepsy is complex. In studies of IGE, the epilepsy and its impact may be important [ 124 ] but there is often dissociation between a good seizure outcome and a poor psychosocial outcome [ 125 ]. A key factor predicting outcome relates to family environment support [ 126 ] but a biological association is supported by the observation that children and adults have an increased risk of psychiatric disturbance, even before the onset of their epilepsy [ 127 , 128 ], and by a broad range of experimental studies [ 129 ]. Interactions between epilepsy and depression may include shared abnormalities in a number of neurotransmitters including 5HT 1A mechanisms [ 130 , 131 ] and via glutamate, where low-dose ketamine, an antiepileptic NMDA antagonist, may have an impact on depression [ 132 ]. These studies illustrate a bidirectional relationship of epilepsy and depression, involving both biological and psychosocial factors.

Table 1

Comorbidities in a nationwide US survey [ 123 ]

	No epilepsy (%)	Epilepsy (%)
Anxiety	13.9	22.4
Depression	25.6	32.5
Bipolar disorder	6.7	14.1
ADHD	5.5	13.2
Sleep disorder/apnea	13.6	19.6
Movement disorder/tremor	4.6	9.3
Migraine	20.6	27.9
Chronic pain	17.7	25.4
Fibromyalgia	7.5	15.4
Neuropathic pain	5.6	8.7
Asthma	16.6	20.7
Diabetes	15.2	15.2
Hypertension	36.7	36.2

A common concern is that antidepressants may increase seizures. The risk of de novo seizures from the use antidepressants is 0.1% for newer drugs and 0.3% from older drugs, e.g. tricyclics [ 133 ]. Exceptions may be maprotiline, bupropion or clomipramine with a higher risk [ 134 ] but overall, those in the treatment arm of antidepressant trials had fewer seizures than those in the placebo arms [ 134 ]. In smaller studies of those with epilepsy at therapeutic doses of antidepressants, many will experience an improvement in their epilepsy [ 135 ]. A recent review has brought together the newer mechanistic evidence, showing that 5HT 1A may mediate a number of actions, which have antiepileptic effects, including increasing GABA activity and reducing inflammatory cytokines and those patients with epilepsy may have reduced PET ligand binding at 5 HT 1A sites [ 136 ]. In a mouse model of sudden unexplained death in epilepsy, drugs acting on 5-HT 3 receptors (fluoxetine, blocked by ondansetron) reduced respiratory arrest in seizures, without affecting the seizures themselves [ 137 ], a further possible mode of benefit of antidepressants in epilepsy. Where possible, it may be appropriate to avoid those antidepressants with pharmacokinetic interactions with AED, such as fluvoxamine, paroxetine and fluoxetine. Hopefully, neurologists can now encourage the use of antidepressants, especially as psychiatric comorbidity is a greater determinant of quality of life than seizure frequency in those with refractory epilepsy [ 138 ].

Antiepileptic drugs and pregnancy

In recent years, information regarding major congenital malformation (MCM) rates has been consolidated in epilepsy and pregnancy registries. AED are divided into those with reasonably quantified risk and those with insufficient data. This becomes self-reinforcing with increased reluctance to prescribe drugs of uncertain risk to those who may conceive. The most recent data from the UK epilepsy and pregnancy register, shows a very clear dose-related effect with valproate risk 5% with <600 mg daily increasing to 11% at over 1000 mg. Carbamazepine at 2% risk when given at <500 mg daily, 3% at 500–1000 mg and 5% at >1000 mg. Lamotrigine had a less steep curve with 2% at <200 mg, increasing to 3.5% over 400 mg daily [ 139 ]. These data are similar to those published from European and US registries [ 140 ]. Oxcarbazepine, not widely used in the UK, appears to have similar low risk to lamotrigine at 2.2% [ 141 ]. The risk for levetiracetam appears similarly low at 0.7% in monotherapy, increasing in polytherapy [ 142 ]. Added to the risk of MCM are concerns over more subtle neurodevelopmental disturbances, including lower IQ, autism and ADHD, which may conceivably arise from exposure to valproate at any stage of pregnancy [ 143 – 146 ]. Although not widely used in pregnant women, topiramate and zonisamide may be associated with significantly lower birthweight [ 147 ]. Recent data have also shown the importance of considering genetic factors in teratogenicity. A family history of abnormalities increases the risk. The risk to a second child, where a first was affected by an AED may be as high as 17–36% [ 148 , 149 ]. Clinicians must also consider the risk to the mother of epilepsy in pregnancy and data suggest a tenfold increase in mortality compared to non-epilepsy controls, largely due to SUDEP [ 150 ].

Epilepsy surgery

Given the low chance of response to medical therapy after the failure of two AED [ 49 ], this is the widely accepted yardstick for defining refractoriness and the appropriateness for consideration of resective epilepsy surgery. The proportion of patients for whom surgery may be successful is not clear, but is estimated as a maximum of around 2% of the total cohort. With an incidence of 0.5%, in the USA and a prevalence of 750,000, this translates to up to 3500 incident cases and 15,000 prevalent cases, in which surgery might be considered. The rate of epilepsy surgery has remained static at around 1500 cases per year [ 151 , 152 ] for over 20 years. The pattern of cases operated may be changing with a reduction in mesial temporal sclerosis [ 153 ], perhaps due to improved outcomes of childhood febrile seizures. At the same time, the outcomes of extratemporal epilepsies are improving with new diagnostic techniques. The mortality of surgery is around 0.1–0.5% [ 151 ], similar to the annual rate of SUDEP in refractory epilepsy [ 154 ], i.e. the mortality of ongoing refractory epilepsy exceeds the post-operative risk after one year. Complication rates have reduced [ 155 ] and are around 3% for major and 7% for minor complications; one of the commonest complications is a visual field defect after temporal lobectomy [ 151 , 156 ]. The treatment is cost-effective in the long term, with sustained remission and close to half of adult patients and 86% of children may be able to stop their AEDs. Two recent studies have found risk factors for seizure recurrence after post-operative drug withdrawal included pre-operative seizure frequency and post-operative EEG abnormalities [ 157 , 158 ]. They also found about one third of those relapsing will not come back under control with re-introduction of medication, especially those with early recurrence, perhaps reflecting a less complete surgical remission.

Health-related quality of life often returns to normal in those who become seizure free [ 159 ]. Negative prognostic factors include high seizure frequency and long duration at baseline [ 160 , 161 ]. Those with lesions such as cavernomas or benign tumours may achieve 77% seizure freedom at two years, even if surgery is undertaken after a long seizure history [ 162 ].

Advances in epilepsy surgery include alternative methods to resective surgery; improvements in techniques of case selection for surgery and neurostimulation techniques.

Radiosurgery for arteriovenous malformations may give excellent outcomes for associated epilepsy and positive prognostic factors have been reported to be presentation with haemorrhage rather than epilepsy and the absence of post-treatment haemorrhage [ 163 – 165 ]. A recent meta-analysis of stereotactic radiosurgery for mesial temporal sclerosis [ 166 ] showed that the total number of patients reported remains low (<200) but that half became seizure free at a median of 14 months after treatment with a complication rate of around 8% (excluding headache which was more common) and rates of visual field defects similar to open surgery. MRI-guided laser thermocoagulation has been undertaken in a few patient with initially promising results. Procedural morbidity is low and patients may be admitted for just one day. It has been suggested as appropriate particularly for older patients. [ 167 – 170 ]. The electrodes inserted for stereotactic EEG recording may also be used to deliver a thermocoagulation induced lesion to the surrounding brain, with a diameter of 4.5–7 mm. This has been undertaken in patients with hypothalamic hamartoma, for whom surgery is difficult and with a high success in remission of the gelastic seizures associated with these lesions [ 171 ]. Early indications are that this may be an approach which can be undertaken in cases of focal cortical dysplasia.

The identification of patients who will benefit from epilepsy surgery relies on the demonstration of a single brain region responsible for the epilepsy, which can be safely resectable. Identification of a responsible lesion has been demonstrated in numerous studies to predict a better outcome [ 172 ]. Even in those where imaging is normal, resection on the basis of an intracranial EEG abnormality is more likely to result in seizure freedom if the resected tissue is pathologically abnormal [ 173 ]. Increasing pre-operative identification of pathology through improved MRI, through higher field strengths up to 7 T in vivo and enhancing 3 T with automated measures of hippocampal volumes potentially gives a greater chance of identifying candidates who may benefit from surgery [ 174 – 176 ]. In those whom structural imaging remains negative, then FDG-PET can aid in the decision making, either in favour of surgery, e.g. in those thought to have non-dominant TLE or against surgery in more complex cases [ 177 , 178 ]. Magnetoencephalography is not widely used [ 179 ], but a recent study demonstrated that if all MEG abnormal areas were resected, prognosis was improved and MEG can be used to target SEEG more successfully [ 180 ]. Tight clustering of MEG abnormalities predicted a better outcome than more dispersed abnormalities. High density EEG source imaging using increased electrode number may also be valuable in predicting the outcome of surgery [ 181 ]. Intravascular stent EEG, shown to be safe in sheep may be a non-invasive method of intracranial EEG recording in the future [ 182 ].

Where resective surgery is not possible, palliative stimulation techniques may be considered. The most established and widely used is vagus nerve stimulation which is safe, with a low risk of complications, such as infection, haematoma and vocal cord palsy [ 183 ]. An analysis from the VNS registry combined with pooled study data totaling 8423 patients [ 184 ] found that responder rate, defined by a 50% seizure reduction, was 47% at 0–14 months and 63% at 24–48 months with seizure free rates rising from 5–10% over the same period. Quality of life measures also improved with VNS [ 185 ], which may relate to seizure reduction, reduced AED load in association with successful antiepileptic treatment or putative effects of VNS on mood [ 186 ]. Responsive stimulation involves a closed circuit of intracranial electrodes with electrical stimuli delivered to the brain according to a seizure detection paradigm. The circuit is often installed following electrode placement in an unsuccessful attempt to identify a surgical target. In 191 patients there was a 37.9% responder rate compared to 17.3% in the sham group. [ 187 ]. Electrodes placed in the thalamus have been associated with a 69% median reduction in seizure frequency and a 35% rate of serious adverse events, including infection in 10% and lead misplacement in 8% [ 188 ]. Other targets under investigation include the nucleus accumbens [ 189 ] and the cerebellum [ 190 ]. Optogenetic methods [ 191 ], successful in animals, have not yet been applied in humans.

A new classification of epilepsies will support the integration of novel aetiological and genetic factors with the existing electroclinical classification and help identify when a single seizure might be considered epilepsy on the basis of an abnormal EEG or imaging. Midazolam IM has emerged as the benzodiazepine of choice in out-of-hospital treatment of status epilepticus and a valid alternative in hospital, but good clinical studies are lacking beyond this early stage. Limbic encephalitis is increasingly diagnosed and primary treatment is immunotherapy rather than AED. The significance of antibodies more generally in epilepsy remains unclear. Most epilepsy treatment remains without a clear evidence base but ethosuximide and valproate have been demonstrated to be the most efficacious AED in absence epilepsy. Perampanel and lacosamide are new drugs which are emerging as treatments for tonic–clonic seizures in generalized epilepsy. A small number of specific genetic epilepsies have allowed personalized treatment in specific cases but this has not yet had broader application. Epileptogenesis is a fertile area of research and everolimus, an inhibitor of the mTor pathway, has demonstrated efficacy in epilepsy associated with TS, showing the clinical potential of this avenue of research for the first time. Epilepsy and pregnancy registers are consolidating data pointing to the use of lamotrigine, levetiracetam, carbamazepine and/or oxcarbazepine as those AED with the lowest risk of major congenital malformations. New evidence has associated topiramate and zonisamide with low birth weight. Clinicians can treat comorbid depression with most modern antidepressants, reassured that there is little evidence of an adverse effect on their patient’s epilepsy. Surgical treatment of epilepsy remains under-utilised and the selection of patients for surgical treatment of epilepsy is becoming more refined with the use of functional imaging to support structural imaging. Alternative ablative treatments are being explored but are not yet widespread. Stimulation techniques other than VNS are areas of research, which remain to find their place.

Overall, recent epilepsy research has started to change our thinking and approach to patients, as we slowly move towards a more rational basis by which to treat this common condition.

Compliance with ethical standards

Conflicts of interest.

Dr. Manford has no conflicts of interest.

High-Field 7T MRI in a drug-resistant paediatric epilepsy cohort: image comparison and radiological outcomes

Find this author on Google Scholar
Find this author on PubMed
Search for this author on this site
ORCID record for Katy Vecchiato
ORCID record for Chiara Casella
For correspondence: [email protected]
Info/History
Preview PDF

Background and Objectives: Epileptogenic lesions in focal epilepsy can be subtle or undetected on conventional brain MRI. Ultra-high field (7T) MRI offers higher spatial resolution, contrast and signal-to-noise ratio compared to conventional imaging systems and has shown promise in the pre-surgical evaluation of adult focal epilepsy. However, the utility of ultra-high field MRI in paediatric focal epilepsy, where malformations of cortical development are more common, is unclear. This study compared 7T to conventional 3T MRI in children with epilepsy by comparing: (i) scan tolerability; (ii) radiological image quality; (iii) lesion yield. Materials and Methods: Children with drug-resistant focal epilepsy and healthy controls were recruited prospectively and imaged at both 3T and 7T. Safety and tolerability during scanning was assessed via a questionnaire. Image quality was evaluated by an expert paediatric neuroradiologist and estimated quantitatively by comparing cortical thickness between field strengths. To assess lesion detection yield of 7T MRI, a multi-disciplinary team jointly reviewed patients' images. Results: 41 patients (8-17 years, mean=12.6 years, 22 male) and 22 healthy controls (8-17 years, mean=11.7 years, 15 male) were recruited. All children completed the scan, with no significant adverse events. Higher discomfort due to dizziness was reported at 7T (p=0.02), with side-effects more frequently noted in younger children (p=0.02). However, both field strengths were generally well-tolerated and side-effects were transient. 7T images had increased inhomogeneity and artefacts compared to those obtained at 3T. Cortical thickness measurements were significantly thinner at 7T (p<0.001). 8/26 (31%) patients had new lesions identified at 7T which were not identified at 3T, influencing the surgical management in 4/26 (15%). Discussion: 7T MRI in children with epilepsy is feasible, well-tolerated and is associated with a 31% improvement in lesion detection rates.

Competing Interest Statement

The authors have declared no competing interest.

Funding Statement

This research was supported by GOSHCC Sparks Grant V4419, King's Health Partners, in part by the Medical Research Council (UK) (grants MR/ K006355/1 and MR/LO11530/1) and Medical Research Council Centre for Neurodevelopmental Disorders, King's College London (MR/N026063/1), and by core funding from the Wellcome EPSRC Centre for Medical Engineering at King's College London [WT203148/Z/16/Z]. J.O.M, K.V, and C.C were funded by a Sir Henry Dale Fellowship jointly by the Wellcome Trust and the Royal Society (206675/Z/17/Z). C.C was also funded by a grant from GOSHCC (VC1421). T.A. was also supported by an MRC Transition Support Award [MR/V036874/1] and Senior Clinical Fellowship [MR/Y009665/1]. M.E was funded by Action Medical Research (GN2835) and the British Paediatric Neurology Association. R.J.P was funded by a Surgeon-Scientist grant by GOSCHCC (VS0221). This research was funded in whole, or in part, by the Wellcome Trust [WT203148/Z/16/Z and 206675/Z/17/Z] and by the National Institute for Health Research (NIHR) Biomedical Research Centre based at Guy's and St Thomas' NHS Foundation Trust and King's College London and/or the NIHR Clinical Research Facility. The views expressed are those of the author(s) and not necessarily those of the NHS, the NIHR or the Department of Health and Social Care.

Author Declarations

I confirm all relevant ethical guidelines have been followed, and any necessary IRB and/or ethics committee approvals have been obtained.

The details of the IRB/oversight body that provided approval or exemption for the research described are given below:

Ethical approval was granted by the UK Health Research Authority and Health and Care Research Wales (ethics ref. 18/LO/1766).

I confirm that all necessary patient/participant consent has been obtained and the appropriate institutional forms have been archived, and that any patient/participant/sample identifiers included were not known to anyone (e.g., hospital staff, patients or participants themselves) outside the research group so cannot be used to identify individuals.

I understand that all clinical trials and any other prospective interventional studies must be registered with an ICMJE-approved registry, such as ClinicalTrials.gov. I confirm that any such study reported in the manuscript has been registered and the trial registration ID is provided (note: if posting a prospective study registered retrospectively, please provide a statement in the trial ID field explaining why the study was not registered in advance).

I have followed all appropriate research reporting guidelines, such as any relevant EQUATOR Network research reporting checklist(s) and other pertinent material, if applicable.

Data Availability

The clinical and neuroimaging data used in the current work are available from the senior author (J.O.M.) on formal request indicating name and affiliation of the researcher as well as a brief description of the intended use for the data. All requests will undergo King's College London-regulated procedure, thus requiring submission of a Material Transfer Agreement. Full preprocessing steps and the code to run the HCP preprocessing pipeline can be found at https://github.com/Washington-University/HCPpipelines. Please also see https://github.com/Washington-University/workbench for the source code for Connectome Workbench. Other code excerpts, information regarding the analysis, or intermediary results can be made available upon request to [email protected].

View the discussion thread.

Thank you for your interest in spreading the word about medRxiv.

NOTE: Your email address is requested solely to identify you as the sender of this article.

Citation Manager Formats

EndNote (tagged)
EndNote 8 (xml)
RefWorks Tagged
Ref Manager
Tweet Widget
Facebook Like
Google Plus One

Subject Area

Radiology and Imaging
Addiction Medicine (342)
Allergy and Immunology (665)
Anesthesia (180)
Cardiovascular Medicine (2625)
Dentistry and Oral Medicine (314)
Dermatology (222)
Emergency Medicine (397)
Endocrinology (including Diabetes Mellitus and Metabolic Disease) (930)
Epidemiology (12175)
Forensic Medicine (10)
Gastroenterology (756)
Genetic and Genomic Medicine (4064)
Geriatric Medicine (385)
Health Economics (676)
Health Informatics (2625)
Health Policy (997)
Health Systems and Quality Improvement (979)
Hematology (360)
HIV/AIDS (845)
Infectious Diseases (except HIV/AIDS) (13659)
Intensive Care and Critical Care Medicine (790)
Medical Education (398)
Medical Ethics (109)
Nephrology (430)
Neurology (3832)
Nursing (209)
Nutrition (570)
Obstetrics and Gynecology (734)
Occupational and Environmental Health (690)
Oncology (2008)
Ophthalmology (581)
Orthopedics (238)
Otolaryngology (304)
Pain Medicine (250)
Palliative Medicine (73)
Pathology (471)
Pediatrics (1107)
Pharmacology and Therapeutics (459)
Primary Care Research (447)
Psychiatry and Clinical Psychology (3400)
Public and Global Health (6499)
Radiology and Imaging (1390)
Rehabilitation Medicine and Physical Therapy (806)
Respiratory Medicine (869)
Rheumatology (400)
Sexual and Reproductive Health (407)
Sports Medicine (338)
Surgery (441)
Toxicology (52)
Transplantation (185)
Urology (165)

Unveiling the epilepsy enigma: an agile and optimal machine learning approach for detecting inter-ictal state from electroencephalogram signals

Original Research
Open access
Published: 23 August 2024

Cite this article

You have full access to this open access article

Shoibolina Kaushik 1 na1 ,
Mamatha Balachandra 2 ,
Diana Olivia ORCID: orcid.org/0000-0003-2934-2090 3 na1 &
Zaid Khan 1 na1

Epilepsy is a chronic neurological disorder characterized by the occurrence of paroxysmal recurrent seizures, which are caused by abnormal electrical activity in the brain. Seizures vary widely in their presentation, depending on the specific region of the brain involved and the extent of the abnormal electrical discharges. The disease can affect cognitive function posing a serious threat to the patients’ lives. Epilepsy causes emotional and behavioral changes, along with sleep disorders and migraines, leading to social isolation and discrimination. Timely administration of medication can cure most cases of epilepsy. However, identifying epileptic patients requires reviewing multiple EEG signal sheets, which can delay disease prediction. Therefore, the aim of our study is to apply simplistic machine learning algorithms that can study the EEG signal data swiftly to identify individuals in seizure, inter-ictal, and normal states, thereby aiding in medical diagnosis. The novelty of this study lies in the utilization of pre-built methods and develop a fast and efficient model that is lightweight and easy to integrate in healthcare to provide relief to epileptic patients. While previous studies have achieved high accuracy, the discussion involving time complexity of their models has been scarce. Given the importance of timely medication in managing epilepsy, it is crucial to consider the runtime of the model rather than solely focusing on accuracy. Therefore, a model that balances both a short runtime (2.9 min) and a satisfactory accuracy (97.46%) has been developed in this project. Integration of this project's findings will catalyze transformative changes within the healthcare industry, enabling healthcare professionals to detect epilepsy at earlier stages and provide timely interventions, ultimately fostering a system that prioritizes precision, innovation, and improved patient outcomes.

Explore related subjects

Artificial Intelligence

Avoid common mistakes on your manuscript.

1 Introduction

Epileptic seizures are a common manifestation of abnormal brain activity caused by an epileptic disorder, which affects the central nervous system. These seizures often come without warning and can result in a loss of awareness, confusion, and unusual behavior, often leading to injuries. Epilepsy can affect people of all ages, but it is most common in young children and the elderly. Surgery and medication are the traditional methods of treating epilepsy, but these treatments are not always effective and may fail to preventseizures.AccordingtotheWorldHealth Organization (WHO), approximately 50 million people worldwide are living with epilepsy, however, with the right treatment, about 70% of epilepsy patients can be freed from the disease [ 1 ]. Early detection allows for more effective treatment and reduces the restrictions on the person's actions. However, predicting seizures before they occur is a difficult task, and researchers are trying to find ways to predict them. This highlights the importance of developing reliable machine learning algorithms for detecting epileptic seizures with better diagnostic accuracy, as it has a significant impact on human life.

Epileptic Seizure Recognition dataset from the UCI ML Repository is utilized in this work since it is a standardized dataset and serves as a benchmark for comparing different approaches adopted by various authors. The data has been relabeled and preprocessed before extracting features and feeding them into the machine learning models. Additionally, parameter tuning has been performed for the identification of the best set of parameters for each method. A total of 22 models were built out of which an accuracy of 97.46% with training time of 2.9 min was yielded by OnevsRestClassifier model with ExtraTreesClassifier estimator. Ultimately, the OneVsRestClassifier with the ExtraTreesClassifier estimator achieves the highest accuracy and is selected as the final model for the three-class classification.

Thus, this research is focused on identifying the optimal machine learning algorithm for the classification of the epileptic seizure dataset. The objective is to significantly enhance the treatment outcomes and minimize the risk of harm by accurately predicting seizures. The analysis will also encompass the examination of the behavior of the algorithms when exposed to modifications in parameters. Thus, this research will bring forth crucial information on the utilization of classification algorithms in the prediction of epileptic seizures and support future endeavors aimed at improving their accuracy.

The remainder of the paper is structured as follows: Section II comprises of an account of the related work done in machine learning for epileptic seizure detection, section III describes the dataset, pre-processing techniques, and the methods used in this research, section IV presents the results obtained, and section V contains conclusion of this study along with scope of future work.

2 Related work

Over the past two decades, there has been significant progress in the field of epilepsy research, particularly in the area of epilepsy disease classification. Electroencephalography is considered the gold standard for diagnosing epilepsy and is a widely used technique for recording and analyzing brain electrical activity [ 2 ]. In recent years, the classification of epilepsy using EEG signals has received increasing attention from the research community, with the aim of developing automated methods for accurate and efficient diagnosis. This research has produced a wealth of knowledge and techniques that can help identify and classify different types of epilepsy based on EEG signals.

A series of studies have been published in a variety of journals on the topic of seizure detection using EEG signals. In the year 2000, A. Petrosian et al. [ 3 ] used signal wavelet decomposition of the EEG signals and combined it with Recurrent Neural Networks (RNN) to predict epileptic seizures. Three years later, T. Gautama et al. [ 4 ] presented a method for the complete characterization of EEG time series signals to enhance signal classification. V. P. Nigam et al. [ 5 ]. Proposed a neural network-based epilepsy detection system for identifying epileptic seizures from EEG data. The authors of Refs. [ 6 , 7 , 8 , 9 ] investigated the classification of EEG signals using wavelet coefficients. In order to examine the sub-bands of EEG signals in terms of the alpha, beta, gamma, theta, and delta, they utilized multiresolution decomposition and an artificial neural network. They combined the adaptive capabilities of a neural network with the quantitative approach of fuzzy logic and the invariant transformation of the probability density function.

A novel form of recurrent neural network was presented for automated seizure detection. An automated diagnostic method for epilepsy disease detection using the Elman network, a special type of RNN was presented by V. Srinivasan et al. [ 10 ] with 99.6% accuracy. Ref [ 11 , 12 ] suggested the application of correlation metrics for EEG signals and real data, whereas a variety of entropy estimators were used to the EEG signals of epileptic and normal participants in Ref [ 13 , 14 , 15 , 16 ] to identify epileptic seizures. E. D. Übeyli et al. [ 17 ] provided an eigenvector feature extraction approach for EEG signal detection utilizing pattern recognition, while K. Polat et al. [ 18 ] introduced a novel hybrid automated identification system for EEG signal classification.

A. Subasi et al. [ 19 ] used different types of component analysis (Principal Component Analysis, Independent Component Analysis, and Linear Discriminant Analysis) to detect epileptic episodes. A classification technique based on multilayer perceptron neural network was proposed by U. Orhan et al. [ 20 ] for epilepsy medications. L. Wang et al. [ 21 ] presented a multidomain feature extraction method with 99.25% accuracy. Time–frequency analysis was used by D. Gajic et al. [ 22 ] for detecting epileptic activity in EEG signals and an accuracy of 98.7% was obtained. D. Wang et al. [ 23 ] proposed basis-based wavelet packet entropy for feature extraction of EEGs for a seizure detection method with about 100% accuracy. P. Fergus et al. [ 24 ] developed an innovative artificial intelligence approach that was employed for automatic epileptic seizure identification utilizing EEG signals which showed a 10% improvement on existing works. 99.38% accuracy was obtained when genetic algorithms and particle swarm optimization techniques were combined with Support Vector Machine (SVM) algorithm by A. Subasi et al. [ 25 ]. The discrete wavelet transform (DWT) and k- nearest neighbor (k-NN) classifiers were postulated for use by A. Sharmila et al. in [ 26 ] to identify seizures with 99% accuracy. E. Alickovic et al. [ 27 ] used convolutional neural networks (CNNs) and supervised machine learning algorithms for epileptic seizure prediction with an accuracy of 99.77%. A maximum sensitivity of 95.8% was achieved when K.C. Hsu et al. [ 28 ] developed hybrid machine learning methods for detecting epileptic seizures. The authors used a genetic algorithm to select the best features and then fed them into the SVM classifier. R. Rosas-Romero et al. [ 29 ] combined functional Near-Infrared Spectroscopy with CNN with an accuracy ranging from 96.9% to 100%, while L. Hussain [ 30 ] proposed robust epileptic seizures detection techniques using machine learning classification combined with different feature extraction strategies with 99.5% accuracy.

Kruskal–Wallis test was used by Sukriti et al. [ 31 ] to select significant features which were then used by classification models, with random forest yielding the highest accuracy of 98.7%. Jana et al. [ 32 ] obtained 99.47% classification accuracy by reducing EEG channels by about 73% before feeding into CNN. Shen et al. [ 33 ] fed the features extracted by DWT and eigenvalues’ algorithms into SVM for classification and RUSBoosted tree Ensemble method for real-time detection; the method yielded an accuracy of 96.38%. An artificial neural network (ANN) integrated with a relaxed local neighbor difference pattern (RLNDiP) technique was used by S. N. J. et al. [ 34 ] with an accuracy of 95.83%. Morteza Ghazali et al. [ 35 ] applied the Modified Binary Salp Swarm Algorithm (MBSSA), which used a backpropagation classification model as Feed-Forward Neural Network and the authors obtained a highest accuracy of 99.45%.

[ 65 ], [ 48 , 50 , 51 , 52 ] Introduced a machine learning approach for epileptic seizure detection using time–frequency features extracted via the tunable-Q wavelet transform from EEG signals. Basha et al. [ 66 ] presented a hybrid approach combining signal processing techniques and machine learning algorithms for epileptic seizure detection using EEG input, achieving improved accuracy. Das and Nanda [ 67 ] proposed a novel multivariate approach utilizing the BCS-WELM algorithm for detecting epileptic seizures, incorporating both signal processing and machine learning techniques. Rani and Chellam [ 68 ] introduced a novel peak signal feature segmentation process for epileptic seizure detection, focusing on identifying peak features in EEG recordings to enhance accuracy. Kumar et al. [ 69 ]. Presented a novel end- to-end approach employing deep learning techniques for epileptic seizure classification directly from scalp EEG data, eliminating the need for manual feature extraction and achieving promising results. The author in [ 70 ] proposes a method that leverages phase space variability analysis of EEG signals to differentiate between various seizure states. By utilizing the ECOC framework, the method aims to improve the classification accuracy of epileptic seizure states, considering the complexity and variability of EEG data. Authors in [ 53 ] have It is recommended to study the most pertinent predictive models to enhance the quality of future research. Additionally, it is advised to create a dedicated dataset focusing on the absence of epilepsy in children to improve the understanding and prediction of this specific type of epileptic seizure.

In the current year, a tuneable-Q wavelet transform and convolutional neural network-based to develop a real- time method with 97.57% accuracy for detecting epileptic seizures is postulated by Shen et al. [ 36 ]. Qiu X et al. [ 37 ] capture spatial correlations and temporal dependencies for epileptic seizure detection using a different attention ResNet-LSTM network(DARLNet) which yields accuracy ranging from 90.17% to 98.87%. It is found that detection of inter-ictal state holds much importance in locating seizure points in brain and thus, diagnosing epilepsy [ 38 ]. Moreover, epilepsy is a disease that can be controlled if detected early. Thus, it is highly beneficial to the patients if their inter-ictal state is detected through the EEG signals. Many researchers have focused on the application of intricate neural networks to tackle the challenges of epileptic state detection and most of such complex algorithms take time to train on the dataset to provide accurate results. Therefore, the main aim of this study is to utilize pre-built methods and develop a fast and efficient model that is lightweight and easy to integrate in healthcare to provide relief to epileptic patients. The Table 1 shows the contributions and the limitations of the literature work.

3 Methodology

This research follows the pathway as shown in Fig. 1 , with its main objective being three-class classification using a set of twenty-two models. Additionally, the models are also utilized for binary and four-class classification. To accomplish this, the original dataset is replicated three times, and each copy is relabeled accordingly for the specific classification task. The relabeled data undergoes further preprocessing before being fed into the models, and the results are subsequently presented in the following section.

Methodology Flowchart

Input: Epileptic Seizure Recognition datasets.

Output: Categorization of Two, Three and Four classes.

Step 1: Preprocess the dataset (Drop Label five instances).

Step 2: Perform Relabeling of the instances for 2, 3,and 4 classes.

Step 3: Perform a data augmentation to rebalance the dataset using SMOTE technique for 2 and 3 classes.

Step 4: For 2,3, and 4 classes extract the features using Lifting-Based Daubechies Wavelet Transform (LDWT).

Step 5: Divide the dataset into training and testing subsets for each classes.

Step 6: Tune the hyperparameters to increase ML performance.

Step 7: Twenty two different ML models are trained.

Step 7.1: Execute a two-class, three-class and four-class classification.

Step 7.2: Extract the evaluation metrics for training and testing datasets.

For three-class classification, OneVsRestClassifier with ExtraTreesClassifier estimator provides the best accuracy. For Binary and Four class classification OneVsRestClassifier with LightGBM performs better.

Step 8: Finally, well performing models mentioned under Step 7.2 are interpreted using the XAI tools. XAI tools namely, Shapley Additive Explanations, Local Interpretable Model-Agonistic Explanations, ELI5, and DALEX are applied.

3.1 Dataset

For this project, Epileptic Seizure Recognition Dataset, a publicly available dataset on the UCI Machine Learning repository is used [ 39 ]. The dataset includes recordings of brain activity for 500 individuals. In total, there are 11,500 instances of data, with each instance containing 178 data points and a corresponding label (y) in the last column, which can take values from 1 to 5.

The label ‘y’ represents the following category of the 178-dimensional input vector as shown in Fig. 2 :

5—"eyes open", indicating that the EEG recording was taken while the patient had their eyes open.

4—"eyes closed", indicating that the EEG recording was taken while the patient had their eyes closed.

3—"non-tumor brain area", indicating that the EEG recording was taken from a region of the brain without a tumor, and the tumor’s location was identified.

2—"tumor location", indicating that the EEG recording was taken from the area where the tumor was located.

1—"seizure activity", indicating that the EEG recording captured seizure activity (Fig. 3 ).

EEG signals of five separate labels

a Flow diagram of relabelling for three-class. b EEG signals plot for three labels

Table 2 depicts the three phases each label in the dataset can be categorized into. These are ‘Seizure’, ‘Inter-ictal’, and ‘Normal’. ‘Seizure’ phase represents abnormal electrical activity in the brain, ‘Inter-ictal’ phase represents the time between seizures when the brain functions in a normal manner without the abnormal electrical activities of seizure, while the ‘Normal’ phase represents absence of seizure activity.

3.2 Pre-processing

The dataset contains no null values, and follows normal distribution. However, the dataset is modified, and the classes are relabeled accordingly for classification, which are discussed in detail in the subsequent subsections.

3.2.1 Three-class classification

It is found that the eyes closed EEG signals are more pronounced than those recorded with eyes open [ 40 ]. Furthermore, it is observed that label 2 (“tumor location”) and label 3 (“non-tumor brain area”) both represent ‘inter-ictal’ state. Following these, label 5 “eyes open” is dropped, and the label 2 and label 3 are combined to form one label. The three classes thus obtained are as follows, and as depicted through Fig. 4 (a) and (b):

0—“seizure”.

1—“inter-ictal”.

2—“normal”.

The modified dataset thus obtained is unbalanced with 2300 instances for seizure, 4600 instances for inter-ictal, and 2300 instances for normal labels. This imbalance is resolved by the application of Synthetic Minority Over- sampling Technique (SMOTE), a technique to generate synthetic samples of the minority class in order to balance the class distribution. SMOTE works by creating synthetic examples of the minority class by interpolating new samples between the existing minority class samples (Fig. 5 ). This is done by identifying the minority class samples that are in the vicinity of each other and then generating synthetic samples along the line connecting them [ 41 ]. The total number of instances increase from 9200 to 13,800 as shown in Table 3 .

a Flow diagram of relabeling for 4-class. b EEG signals plot for four labels

3.2.2 Binary classification

For binary classification, the label ‘1’ is classified against combined labels ‘2, 3, 4, 5’. Following are the two classes as depicted through Fig. 6 a, b.

a Flow diagram of relabeling for binary classification. b EEG signals plot for two labels

0—“non-seizure”.

1—“seizure”.

The relabeled dataset is unbalanced, with 9200 for ‘non- seizure’ category and 2300 for ‘seizure’ as shown in Table 3 . SMOTE technique is applied to this relabeled dataset to make it balanced. Upon balancing the dataset, the total number of instances become 18,400 (Fig. 7 ).

Visual Representation of ExtraTreesClassifier

3.2.3 Four class classification

After dropping label ‘5’ (denoting the eyes-open EEG signal readings), the modified dataset is left with four classes which are relabeled as follows (refer Fig. 8 a, b).

a Flow diagram of relabeling for 4-class classification. b EEG signals plot for four labels

1—“tumor detected”.

2—“non-tumor area”.

3—“normal”.

The dataset obtained after regrouping the labels contains equal number of instances for every class, hence the dataset is balanced as depicted in Fig. 9 .

Data distribution after grouping labels into four classes

3.2.4 Feature importance

Feature selection is utilized for choosing the essential attributes. Seven feature selection methods have been compared in this study. Figure describes the feature selection methods, the top 70 features chosen by each technique and the names of the features chosen. The figure shows that the Whale optimization algorithm has chosen most of the features. Most of the features are selected by more than four algorithms.

3.3 Algorithms

After regrouping the labels, the modified dataset obtained is subjected to feature extraction using Lifting-Based Daubechies Wavelet Transform (LDWT). LDWT is a fast and memory-efficient method for decomposing signals or images into multi-resolution representations. It uses a series of simple predict and update steps to compute wavelet coefficients, reducing the number of operations required. The resulting transform provides approximation and detail coefficients at different levels, representing the low and high- frequency components, respectively [ 42 ]. LDWT offers computational efficiency and numerical stability, making it a valuable tool for signal analysis, data compression, and feature extraction.

The data is then split into 80% training data, which is fed into the models, and 20% test data, which is used for performance evaluation of the models.

Feature Extraction using LDWT

A total of twenty-two models were developed in this project, out of which the best models yielded from this study is Scikit-learn’s OneVsRestClassifier with a variety of estimator models. OneVsRestClassifier trains a single classifier to solve a multi-class classification problem by dividing it into multiple binary classification tasks. In this method, for each class, a binary classification model is trained to distinguish it from all other classes. In essence, it converts the multi-class classification problem into a set of binary classification problems. The estimator model is ExtraTreesClassifier in case of three-class classification and Light Gradient Boosting Machine (LightGBM) for binary and four-class classification.

3.3.1 OneVsRestClassifier with ExtraTreesClassifier estimator for three-class classification

ExtraTreesClassifier is a variant of the Random Forest algorithm and is used for classification tasks. The general working of the model consists of three main steps: (i) Bagging, (ii) Feature selection, and (iii) Voting. Bagging, also known as bootstrap aggregating, is a technique through using which the model builds a large number of decision trees, where each decision tree is trained on a randomly sampled subset of the training data. After bagging, the model selects a random subset of features for every decision tree at each node in the tree. This is feature selection, also known as attribute bagging. Bootstrap aggregating and feature selection prevents ExtraTreesClassifier from overfitting, while the majority voting method makes it more robust to noise, while also providing high accuracy results in various classification tasks, especially in high dimensional feature spaces. Finally, the model combines the results of all the decision trees by using majority voting to predict the final class label for a given input. Figure 10 shows the working of ExtraTreesClassifier algorithm in which Ii represents input for ith decision tree, Ti indicates ith decision tree, T* depicts final decision tree.

For three-class classification, OneVsRestClassifier gives the best accuracy using ExtraTreesClassifier as the estimator model. Algorithm 2 highlights the pathway of the best model for classifying of the dataset into “seizure”, “inter-ictal”, and “normal” classes.

OneVsRestClassifier with ExtraTreesClassifier estimator for three-class classification

3.3.2 OneVsRestClassifier with LightGBM estimator for binary classification and four-class classification

Boosting algorithms combine weak models to create a powerful predictive model. These algorithms iteratively train weak learners, such as decision trees, by adjusting instance weights based on previous models' performance. The ensemble of weak models with weighted predictions forms a strong model that surpasses the individual learners. They effectively capture complex patterns in data thus improving predictive accuracy. LightGBM, a type of boosting algorithm, uses a novel technique called Gradient-based One- Side Sampling (GOSS) to perform feature selection during the training process [ 43 ]. GOSS selects the most informative data points for training the model, while discarding the less informative ones which allows LightGBM to be much faster and more efficient than other gradient boosting algorithms. Once the informative data points are selected, LightGBM builds a decision tree based on these data points using a leaf- wise approach.

LightGBM when used as the base model for binary classification and four-class classification provides the best accuracy. The algorithm is as outlined in Algorithm 3:

OneVsRestClassifier with LightGBM as estimator model for binary classification and four-class classification

4 Results and discussion

The results are based on the performance evaluation of twenty-two models developed. These include Convolutional Neural Network (CNN), Artificial Neural Network (ANN), Long Short-Term Memory Network (LSTM), Bidirectional Long Short-Term Memory Network (BiLSTM), Categorical Boosting (CatBoost), Extreme Gradient Boosting (XGBoost), Light Gradient Boosting Machine (LightGBM), ClassifierChain, and MultiOutputClassifer.

4.1 Three-class classification results

A summary of all the twenty-two models’ performance is provided by Table 4 . The table also depicts if feature extraction methods, viz. LDWT and Principal Component Analysis (PCA) is applied during pre-processing. PCA is a dimensionality reduction technique in machine learning that allows the extraction of essential information from high- dimensional datasets by transforming the original features into uncorrelated variables called principal components [ 44 ].

PCA helps reduce the number of input features, removes noise, aids in visualization, and serves as a preprocessing step for other algorithms.

A detailed report on the best model for three-class classification is given by Table 5 , which provides the precision, recall, and f1-score for every label (0—“seizure”, 1—“inter-ictal”, 2—“normal”). For “inter-ictal” detections, the model provides reliable results that can be used in epilepsy diagnosis.

4.2 Binary classification and four-class classification

A summary of all the models’ performance for binary and four-class classifications is provided by Table 6 .

4.3 Explainable artificial intelligence (XAI)

In this study, we want to build classification pipelines and give the predictions we make some meaning [ 47 ]. We used a variety of XAI tools to explain different classifiers, including Shapley Additive Explanations, Local Interpretable Model- Agonistic Explanations, ELI5, and DALEX. Four tree-based feature significance graphs and medical literature are then used to validate these XAI approaches.

4.3.1 Shapley additive explanations (SHAP)

A popular tool for game theoretic classifier explanation is SHAP. Lundberg et al. [ 45 ]. proposed this mathematical model. Each feature's Shapley values are assessed according to how well they contribute to a prediction. The SHAP Beeswarm plot obtained for the OneVsRestClassifier with ExtraTreesClassifier estimator for three class classifier is shown in Fig. 11 . A comprehensive global interpretation summary for the OneVsRestClassifier with ExtraTreesClassifier is provided by the beeswarm plot. One data point is represented by each dot on the plot. In order of decreasing relevance, the feature names are positioned along the y-axis. Along the x-axis, the SHAP values are displayed. The feature values are indicated by the color gradient. The higher the feature value of a dot on the plot, the redder it is. In this plot, it can be observed that the feature X79 has the most significant contribution to the output class Inter-ictal and X152 feature for Seizure class.

Features chosen by the various algorithms

This positive relation indicates that a higher value of these features could increase the patient’s probability of being classified as Inter-ictal and Seizure respectively.

4.3.2 Local interpretable model-agonistic explanations (LIME)

Another tool for classifier explanations is LIME. Individual forecasts can be given interpretations using this technique. Ribeiro [ 46 ] advocated this approach. This algorithm assesses the results of altering the inputs to produce predictions. LIME gave an explanation for the local predictions made by the OneVsRestClassifier with ExtraTreesClassifier. This tool displays a plot of the contribution of each feature to a prediction as shown in Fig. 12 .

Beeswarm SHAP plot for explaining the OneVsRestClassifier with ExtraTreesClassifier for three class model: a Inter-ictal class b Seizure class c Normal class

It can be seen that the most significant feature X79 has contributed to raising the likelihood that the patient would be classified as Inter-ictal, similarly X152 feature contributed to classifying the patient as a Seizure.

4.3.3 Explain like I’m 5 (ELI5 tool)

This Python module helps with both debugging the machine learning classifier and explaining pre dictions. Tree-based classifiers are supported by this tool. OneVsRestClassifier with ExtraTreesClassifier was explained in this study using ELI5. Both regional and global explanations are possible with ELI5 [ 47 ]. The tabular interpretation of OneVsRestClassifier with ExtraTreesClassifier is shown in Fig. 13 . The highest contributing feature is shown on the first row of a list of features in Fig. 13 . When building a model, the X79 feature received the most weight and were chosen as the root node. According to how much each feature contributed to the forecast, the weights of each feature are shown from highest to lowest. Figure 14 ELI5 explanation plots: Global explanation for three class model OneVsRestClassifier with ExtraTreesClassifier.

Local Interpretable Model-Agonist for three class model OneVsRestClassifier with ExtraTreesClassifier: a Normal b Inter-ictal, Seizure

Beeswarm SHAP plot for explaining the four class OneVsRestClassifier with LightGBM estimator: a Seizure b Non-tumor area c Tumor detected(class0) d Normal(class0)

In four class classification, across all classes namely Seizure, Tumor Detected, Non-Tumor Area, and Normal, which is best used by OnevsRestClassifier with LightGBM. SHAP consistently identified the same set of top features for positive detection (class 1) as well as its absence (class 0). This is in line with the foundational principles of SHAP, which leans on game theory to allocate an influence value to each feature, ensuring explanations adhere to the model's overarching behavior as shown in Fig. 14 . When compared with LIME, certain harmonies became evident. As shown in Fig. 15 , for the Seizure class, both methodologies underscored the importance of feature X141 and X95. This synchronization continued in the class 0 of Tumor Detected class, with both SHAP and LIME highlighting features like X123, X87, X118, and X34. However, there were instances of unique insights. In the Non-Tumor Area classification, while SHAP unveiled a diverse set of influential features, LIME honed in on the prominence of feature X82 which was seen in SHAP as well. In the class 0 of Normal Area Classification X141 and X113 were points to be matched between SHAP and LIME.

Local Interpretable Model-Agonist for four class OneVsRestClassifier with LightGBM estimator: a Normal b Seizure, Tumor detected c Non-tumor area

For four class classification as shown in Fig. 16 , ELI5's features such as X118, X 141,X 69,X 25, X32, X164, X34, X154, X133, X67, X152, X19, X163, X1, X36, X157, X136, X95, X79, X73, and X147 exhibit a pronounced overlap with SHAP's features, especially for the classes "seizure", "non tumor area", and "normal". The overlap with SHAP's "seizure" and "non tumor area" class (both class 1) is particularly notable. LIME also echoes some of ELI5's findings, especially for features X141, X113, and X95 in the "seizure" and "not normal" categories, suggesting these features have significant importance. Across the board, ELI5's identified features seem to align more with SHAP and LIME's class 1 outputs ("seizure", "tumor detected", "non tumor area", and "normal") than with their inverse or "NOT" classes. The Feature X82 was also notable for 4 class SHAP Non-tumorarea, Tumor detected(class0), Normal(class0), and 4 class lime for non tumor class. The Feature X79, is highly influential in the 3-class model for predicting the Inter- ictal and Seizure classes, as indicated by SHAP. Also noted in ELI5's analysis for its significant contribution. Also in LIME for Inter-Ictal class.

ELI5 explanation plots: global explanation for four class OneVsRestClassifier with LightGBM estimator

4.3.4 Descriptive mAchine learning explanations (DALEX)

DALEX [ 49 ], an acronym for Descriptive mAchine Learning EXplanations, provides two critical types of interpretability tools: the Model Parts Plot and the Break Down Plot, each serving distinct purposes in the realm of machine learning model interpretation.

4.3.4.1 Model parts plot (global interpretation)

This plot, also known as a feature importance or drop-out loss plot, offers a global view of the model's behavior. It quantifies and visualizes the impact of each feature on the model’s predictive performance. By iteratively removing or permuting each feature and measuring the resultant loss in model accuracy, it assigns an importance score to each feature. The plot displays these scores, allowing users to quickly identify which features are driving the model's predictions. This global perspective is invaluable for understanding the overall behavior of complex models, highlighting the features that significantly contribute to the model's accuracy.

4.3.4.2 3-class model using one Vs rest classifier with ExtraTreesClassifier

Plot Interpretation: In this plot as shown in Fig. 17 , the variables are again ranked by dropout loss. Here, X146 is the most important feature, followedbyX104, X152, X127, X33, X172, X31, X173, X130, and X66. The scale of the dropout loss is more substantial in this model, suggesting that the features have a more pronounced effect on model performance. Feature Importance: The features listed are those that, when dropped, negatively impact the model's accuracy. The higher the dropout loss, the more critical the feature is to the model's predictions.

Descriptive mAchine Learning Explanations (Model Parts Plot (Global Interpretation): 3 class onevsrest with extratrees classifier

4.3.4.3 4-class model using OneVsRestClassifier with LightGBM plot interpretation

The plot shown in Fig. 18 , the variable X146 has the highest dropout loss, meaning that removing this feature causes the most significant decrease in model performance. This is followed by X71, X141, X19, X15, X34, X13, X139, X137, X52. The dropout loss is relatively close among the top variables, indicating that several features contribute similarly to model performance. Feature Importance: As with the 4-class model, the features shown are key to the model's accuracy. The greater the loss, the higher the importance. It's notable that X146 is consistently at the top in both models, underscoring its significance.

Descriptive mAchine Learning Explanations(Model Parts Plot (Global Interpretation): 4 class onevsrest with lightgbm classifier

4.4 Comparative analysis

When comparing both models, the following observations can be made: The feature X146 appears to be the most critical feature in both models, indicating its strong relationship with the target variable in epileptic seizure recognition. There is a difference in the ranking and importance of other features between the two models, which suggests that the underlying algorithms may be focusing on different patterns within the data. The dropout loss values in the 3-class model are generally higher, which could imply that the features in this model have a more substantial impact on performance, or it could reflect differences in scale between the two problems. Final List of Outperformed Features. To determine a final list of outperformed features for epileptic seizure recognition, one should consider the following:

Feature that appear in both lists (X146) is likely to be highly relevant.

Break down plot (local interpretation):

In contrast to the global insights provided by the Model Parts Plot, the Break Down Plot offers a granular, instance- level analysis. It dissects a single prediction, detailing how each feature's value for that specific instance contributes to the final prediction. The plot starts from the model's baseline prediction and adds or subtracts the impact of each feature, culminating in the final predicted output. This step-by-step breakdown provides an in-depth understanding of the model's reasoning for a particular instance, making it an essential tool for local interpretability. It allows users to trace the model's decision-making process, uncovering how individual features sway the model towards or away from a particular outcome. In essence, DALEX's interpretability plots cater to both macro and micro-level insights into machine learning models.

While the Model Parts Plot offers a bird's-eye view of feature significance across the model, the Break Down Plot dives into the intricate details of individual predictions, providing a comprehensive suite for interpreting and validating complex machine learning models in various scenarios.

4.5 ModelAi

4.5.1 normal class.

For the Normal class, the features contribute as shown in Figure 19 (A) as follows. The intercept starts with a negative contribution, suggesting that without considering any features, the model is inclined not to predict the Normal class. Most features listed have a very slight negative contribution, which means their presence slightly decreases the likelihood of predicting the Normal class. The 'all other factors' have a large negative impact, further reducing the likelihood of a Normal prediction. This could indicate that the model has learned a specific pattern for the Normal class that is not represented by the individual features shown contributions, implying that higher values for these features make a Seizure prediction less likely.

Descriptive mAchine learning explanations [Break Down Plot (Local Interpretation)]: 3 class onevsrest with extratrees classifier. A normal B inter-ictal C seizure

'All other factors' combined have a negative contribution, suggesting that the majority of features not individually listed are collectively reducing the likelihood of a Seizure prediction.

As shown in Fig. 19 (B), the positive contributions indicate how much each feature moves the prediction away from the baseline (the intercept) towards predicting an Inter-ictal state: The intercept is set at around 0.337, which is the baseline prediction value before accounting for the effect of any features. Features like X91 and X17 have a strong negative influence on the prediction, meaning that as their values increase, the likelihood of predicting an Inter-ictal state decreases. Conversely, features like X28 have a positive effect, increasing the likelihood of an Inter-ictal prediction. The 'all other factors' bar suggests that when combining the effects of all other features not listed, the overall contribution is strongly positive, indicating that many smaller contributions push the model towards an Inter-ictal prediction.

4.5.2 Seizure class

The Seizure class breakdown is interpreted as shown in Fig. 19 (C) as follows:

The baseline prediction for the Seizure class is higher than for the Normal class, as indicated by the intercept.

Several features have positive contributions, but these are relatively small. Features X95, X4, and X7 have negative.

4.5.3 Technical insights

The breakdown plots illustrate the additive nature of the model's predictions, where the final prediction is a sum of the baseline (intercept) and individual feature contributions.

Features with larger absolute contributions are more influential in the model's decision-making process. In this context, features with large positive contributions are strong indicators for their respective classes. The intercept represents the average prediction when all features are at their baseline or average values. If the intercept is high for a particular class, it suggests that the class is the default or most common prediction without any further information. The 'all other factors' bar accounts for the cumulative contribution of all features not individually listed in the breakdown plot. A large positive or negative bar indicates that many small contributions are influencing the prediction in one direction.

The breakdown plots for the 4-class classification using OneVsRest with LightGBM highlight the contribution of individual features towards the prediction of four different classes: Normal, Seizure, Tumor Detected, and Non-tumor Area.

4.6.1 Normal class

Intercept: The baseline prediction is positive, indicating a predisposition towards the Normal class before considering the feature effects as shown in Fig. 20 (A). Positive Contributions: Features such as X4, X130, X141, X60, X56 contribute positively towards predicting the Normal class. This means that higher values of these features make a Normal class prediction more likely.

Descriptive mAchine learning explanations [Break Down Plot (Local Interpretation)] 4 class onevsrest with lightgbm classifier. A normal B tumor detected C seizure D non-tumor

Negative Contributions: Features like X8, X139 and X134 negatively influence the prediction, suggesting that their higher values decrease the likelihood of the Normal class.

4.6.2 Seizure class

Intercept: As shown in Fig. 20 (B) the prediction starts off with a positive bias towards the Seizure class. Positive Contributions: Features such as X4, X1, X15, X66, and X130 have the most substantial positive impact, increasing the prediction probability for the Seizure class. Negative Contributions: Few features like X126 and X34 decrease the likelihood of predicting the Seizure class.

4.6.3 Tumor detected class

Intercept: As shown in Fig. 20 (C) there's a positive baseline prediction for the Tumor Detected class. Positive Contributions: Features X119, X96, X158, and X52 notably push the prediction towards a Tumor Detected outcome. Negative Contributions: X137 has a significant negative impact, whereas X101 and X119 have slight negative influences.

4.6.4 Non-tumor area class

Intercept: As shown in Fig. 20 (D) the baseline prediction is neutral. Positive Contributions: Features such as X178, X25, and X17 lead to a higher probability of predicting the Non-tumor Area class.

Negative Contributions: A large negative contribution from X170 and X11 suggests these features strongly reduce the likelihood of this class being predicted.

Feature X104, X127, X33, X172, X31, X173, X130, and X66: These features were also noted in DALEX's Model Parts Plot for the 3-class model. Their importance is highlighted by their impact on the dropout loss when removed, affecting the model's accuracy. The Feature X71, X19, X15, X13, X139, X137, and X52: Identified in the DALEX's Model Parts Plot for the 4-class model using OneVsRestClassifier with LightGBM. These features, like X146, were found to cause a significant decrease in model performance when removed, highlighting their importance.

Feature X146: Emerged as a top feature in both 3-class and 4-class models using different classifiers (ExtraTreesClassifier and LightGBM). Consistently identified by DALEX, indicating its strong relationship with the target variable in epileptic seizure recognition.

4.7 Feature X152

Identified by SHAP in the 3-class model as crucial for predicting the Seizure class.

Similar findings in LIME and ELI5 analysis, reinforcing its importance. Also found in DALEX 3 Class Global Interpretation plot as significant.

4.8 Feature X141 and X95

Both features were highlighted in the 4-class model analysis using SHAP and LIME for the Seizure class. ELI5's findings in the 4-class model also echoed the significance of these features.

X141 was also found in DALEX (Model Parts Plot (Global Interpretation): 4 class OneVsRest with LightGBM classifier and in DALEX (Break Down Plot (Local Interpretation)) 4 Class OneVsRest with LightGBM classifier of Normal class contributing positively.

X95 was found in Dalex (Break Down Plot (Local Interpretation)): 3 class OneVsRest with extratrees Classifier of Seizure class contributive negatively.

4.9 Feature X34 and X113

Notable for their contributions in the 4-class model, particularly in the context of Seizure and Normal classes. These features were underscored by both SHAP and LIME, as well as ELI5 for X34. X113 was also found in DALEX (Model Parts Plot (Global Interpretation):4 class OneVsRest with LightGBM classifier.

5 Discussions

During the development of the models, upon implementation of binary classification, the LightGBM model with the OneVsRestClassifier approach demonstrated the highest accuracy. Naturally, it was anticipated that the same model would also exhibit the best accuracy when applied to the three-class classification problem. However, the ExtraTreesClassifier estimator outperformed the LightGBM model in terms of accuracy for the three-class classification task.

Another expectation was that reducing the number of features using PCA would enhance the efficiency of the models. However, much to the contrary, the best-performing models for both binary and three-class classification did not make use of PCA. These deviations can be attributed to the preference of these models to train on a larger amount of data. It appears that they achieve more accurate results when trained on the maximum available data rather than utilizing feature reduction techniques such as PCA.

Such unexpected results indicate that the specific characteristics and complexities of the dataset and classification task influenced the performance of the models. It emphasizes the importance of thorough experimentation and exploration to uncover the most effective approaches for a given problem, rather than relying solely on initial assumptions or expectations. Further various Explainable Artificial Intelligence tools, SHAP, LIME, ELI5, and DALEX explained the high-performing models for three and four class classifiers. It was observed that all XAI tools results align with each other in three and four class classifiers.

6 Conclusion

The novelty of this study lies in the application of simple yet effective classification algorithms that provide results within a very short time. The motivation is to aid the healthcare providers in saving time by utilizing a simpler approach, as opposed to utilizing intricate algorithms that consume more time. The model proposed in this paper provides an accuracy of 97.46% with training time as less as 2.9 min (174.28 s). We obtained valuable insights into the rationale behind the best-performing models with explainable AI tools such as SHAP, LIME, ELI5, and DALEX.

For future work, there are several areas of focus to improve the proposed model. Firstly, efforts can be made to reduce the model's runtime even further. While the current runtime of 2.9 min is relatively low, additional optimizations can be explored to make it more efficient and responsive. Another aspect to address is the enhancement of the parameter set used in the model. By fine-tuning the parameters, it is possible to achieve better accuracy results. Moreover, the proposed model's accuracy can be further improved by exploring alternative feature extraction techniques and incorporating them into the classification process. By considering different feature sets or incorporating domain-specific knowledge, it may be possible to capture more relevant information and enhance the model's discriminatory power.

Since the ultimate goal is to integrate this model into the healthcare system, close collaboration with medical professionals and system developers is required to ensure the model's compatibility, reliability, and security within the existing healthcare infrastructure. This successful integration would guarantee epileptic patients prompt medical attention, thereby reducing the potential risks and consequences associated with delayed treatment. Therefore, the outcome of this paper has the potential to make a significant impact on the management of epilepsy and improve the quality of life for individuals living with the condition.

In future the study can be integrated with other physiological signals, such as ECG and EMG along with EEG to enhance the accuracy of the model. The model can be integrated with a real-time monitoring system comprising of wearable devices to allow early intervention and personalized treatment. Finally, data augmentation and transfer learning techniques can be included to enhance the generalization technique of the seizure detection system.

World health organization, Epilepsy www.who.int .

Rosenow F, Klein KM, Hamer HM (2015) Non-invasive EEG evaluation in epilepsy diagnosis. Expert Rev Neurother 15(4):425–444

Article Google Scholar

Petrosian A, Prokhorov D, Homan R, Dasheiff R, Wunsch D (2000) Recurrent neural network based prediction of epileptic seizures in intra- and extracranial EEG. Neurocomputing 30(1–4):201–218

Gautama T, Mandic DP, Van Hulle MM (2003) Indications of nonlinear structures in brain electrical activity. Phys Rev E 67:4. https://doi.org/10.1103/physreve.67.046204

Nigam VP, Graupe D (2004) A neural-network-based detection of epilepsy. Neurol Res 26(1):55–60. https://doi.org/10.1179/016164104773026534

Güler İ, Übeyli ED (2005) Adaptive neuro-fuzzy inference system for classification of EEG signals using wavelet coefficients. J Neurosci Methods 148(2):113–121. https://doi.org/10.1016/j.jneumeth.2005.04.013

Subasi A (2007) EEG signal classification using wavelet feature extraction and a mixture of expert model. Expert Syst Appl 32(4):1084–1093. https://doi.org/10.1016/j.eswa.2006.02.005

Adeli H, Ghosh-Dastidar S, Dadmehr N (2007) A wavelet-chaos methodology for analysis of EEGs and EEG subbands to detect seizure and epilepsy. IEEE Trans Biomed Eng 54(2):205–211. https://doi.org/10.1109/tbme.2006.886855

Guo L, Rivero D, Dorado J, Rabuñal JR, Pazos A (2010) Automatic epileptic seizure detection in EEGs based on line length feature and artificial neural networks. J Neurosci Methods 191(1):101–109. https://doi.org/10.1016/j.jneumeth.2010.05.020

Srinivasan V, Eswaran C, Sriraam N (2005) Artificial neural network based epileptic detection using time-domain and frequency- domain features. J Med Syst 29(6):647–660. https://doi.org/10.1007/s10916-005-6133-1

Kannathal N, Acharya UR, Lim CM, Sadasivan PK (2005) Characterization of EEG—A comparative study. Comput Methods Programs Biomed 80(1):17–23. https://doi.org/10.1016/j.cmpb.2005.06.005

Harikrishnan KP, Misra R, Ambika G, Kembhavi AK (2006) A non-subjective approach to the GP algorithm for analysing noisy time series. Physica D 215(2):137–145. https://doi.org/10.1016/j.physd.2006.01.027

Article MathSciNet Google Scholar

Kannathal N, Choo ML, Acharya UR, Sadasivan PK (2005) Entropies for detection of epilepsy in EEG. Comput Methods Programs Biomed 80(3):187–194. https://doi.org/10.1016/j.cmpb.2005.06.012

Srinivasan V, Eswaran C, Sriraam N (2007) Approximate entropy- based epileptic EEG detection using artificial neural networks. IEEE Trans Inf Technol Biomed 11(3):288–295. https://doi.org/10.1109/titb.2006.884369

Nicolaou N, Georgiou J (2012) Detection of epileptic electroencephalogram based on permutation entropy and support vector machines. Expert Syst Appl 39(1):202–209. https://doi.org/10.1016/j.eswa.2011.07.008

Kumar Y, Dewal ML, Anand RS (2012) Epileptic seizures detection in EEG using DWT-based ApEn and artificial neural network. SIViP 8(7):1323–1334. https://doi.org/10.1007/s11760-012-0362-9

Übeyli ED, Güler İ (2007) Features extracted by eigenvector methods for detecting variability of EEG signals. Pattern Recogn Lett 28(5):592–603. https://doi.org/10.1016/j.patrec.2006.10.004

Polat K, Güneş S (2008) Artificial immune recognition system with fuzzy resource allocation mechanism classifier, principal component analysis and FFT method based new hybrid automated identification system for classification of EEG signals. Expert Syst Appl 34(3):2039–2048. https://doi.org/10.1016/j.eswa.2007.02.009

Subasi M, Ismail G (2010) EEG signal classification using PCA, ICA, LDA and support vector machines. Exp Syst Applicat 37(12):8659–8666. https://doi.org/10.1016/j.eswa.2010.06.065

Orhan U, Hekim M, Ozer M (2011) EEG signals classification using the K-means clustering and a multilayer perceptron neural network model. Expert Syst Appl 38(10):13475–13481. https://doi.org/10.1016/j.eswa.2011.04.149

Wang L, Xue W, Yang L, Luo M, Huang J, Cui W, Huang C (2017) Automatic epileptic seizure detection in EEG signals using multi- domain feature extraction and nonlinear analysis. Entropy 19(6):222

D Gajic, Z Djurovic, J Gligorijevic, S Di Gennaro, and I Savic- Gajic (2015) Detection of epileptiform activity in EEG signals based on time-frequency and non-linear analysis, Front Comput Neurosci 9

Wang D, Miao D, Xie C (2011) Best basis-based wavelet packet entropy feature extraction and hierarchical EEG classification for epileptic detection. Expert Syst Appl. https://doi.org/10.1016/j.eswa.2011.05.096

Fergus P, Hignett D, Hussain A, Al-Jumeily D, Abdel-Aziz K (2015) Automatic epileptic seizure detection using scalp EEG and advanced artificial intelligence techniques. Biomed Res Int. https://doi.org/10.1155/2015/986736

Subasi JK, Abdullah Canbaz M (2017) Epileptic seizure detection using hybrid machine learning methods. Neural Comp Applicat 31(1):317–325. https://doi.org/10.1007/s00521-017-3003-y

Sharmila Madan S, Srivastava K (2018) Epilepsy detection using DWT based hurst exponent and SVM, K-NN classifiers. Serb J Exp Clin Res 19(4):311–319. https://doi.org/10.1515/sjecr-2017-0043

Alickovic E, Kevric J, Subasi A (2018) Performance evaluation of empirical mode decomposition, discrete wavelet transform, and wavelet packed decomposition for automated epileptic seizure detection and prediction. Biomed Signal Process Control 39:94–102. https://doi.org/10.1016/j.bspc.2017.07.022

Hsu KC, Yu SN (2010) Detection of seizures in EEG using subband nonlinear parameters and genetic algorithm. Comput Biol Med 40(10):823–830. https://doi.org/10.1016/j.compbiomed.2010.08.005

Rosas-Romero R et al (2019) Prediction of epileptic seizures with convolutional neural networks and functional near-infrared spectroscopy signals. Comput Biol Med. https://doi.org/10.1016/j.compbiomed.2019.103355

Hussain L (2018) Detecting epileptic seizure with different feature extracting strategies using robust machine learning classification techniques by applying advance parameter optimization approach. Cogn Neurodyn 12(3):271–294. https://doi.org/10.1007/s11571-018-9477-1

Sukriti MC, Mitra D (2021) Epilepsy seizure detection using kurtosis based VMD’s parameters selection and bandwidth features. Biomed Sign Proc Cont. https://doi.org/10.1016/j.bspc.2020.102255

Jana R, Mukherjee I (2021) Deep learning based efficient epileptic seizure prediction with EEG channel optimization. Biomed Signal Process Control. https://doi.org/10.1016/j.bspc.2021.102767

Shen M, Wen P, Song B, Li Y (2022) An EEG based real-time epilepsy seizure detection approach using discrete wavelet transform and machine learning methods. Biomed Signal Process Control. https://doi.org/10.1016/j.bspc.2022.103820

SNJ, SMSP, and TGS (2022) EEG-based classification of normal and seizure types using relaxed local neighbour difference pattern and artificial neural network, Knowl Based Syst 249: 108508 https://doi.org/10.1016/j.knosys.2022.108508

Morteza Ghazali S, Alizadeh M, Mazloum J, Baleghi Y (2022) Modified binary salp swarm algorithm in EEG signal classification for epilepsy seizure detection. Biomed Signal Process Control. https://doi.org/10.1016/j.bspc.2022.103858

Shen M, Wen P, Song B, Li Y (2023) Real-time epilepsy seizure detection based on EEG using tunable-Q wavelet transform and convolutional neural network. Biomed Signal Process Control. https://doi.org/10.1016/j.bspc.2022.104566

Qiu X, Yan F, Liu H (2023) A difference attention ResNet-LSTM network for epileptic seizure detection using EEG signal. Biomed Signal Process Control. https://doi.org/10.1016/j.bspc.2023.104652

Mann EO, Mody I (2009) GABA | synchrony through GABAergic inhibition. Encyclop Basic Epilep Res. https://doi.org/10.1016/b978-012373961-2.00131-4

Harun-Ur-Rashid, “Epileptic seizure recognition,” Kaggle. Available: https://www.kaggle.com/datasets/harunshimanto/epileptic-seizure - recognition

Barry RJ, Clarke AR, Johnstone SJ, Magee CA, Rushby JA (2007) EEG differences between eyes-closed and eyes-open resting conditions. Clin Neurophysiol 118(12):2765–2773. https://doi.org/10.1016/j.clinph.2007.07.028

Chawla NV, Bowyer KW, Hall LJ, Kegelmeyer WP (2002) SMOTE: synthetic minority over-sampling technique. J Artific Intell Res 16:321–357. https://doi.org/10.1613/jair.953

Sweldens W (1996) The lifting scheme: a custom-design construction of biorthogonal wavelets. Appl Comput Harmon Anal 3(2):186–200. https://doi.org/10.1006/acha.1996.0015

G Ke, Q Meng, T Finley, T Wang, W Chen and Q Ye (2017) LightGBM: a highly efficient gradient boosting decision tree, in Proceedings of the 31st conference on neural information processing systems (NIPS 2017), Long Beach, CA, USA 3146–3154

Pearson K (1901) On lines and planes of closest fit to systems of points in space. Lond Ed Dub Philosop Mag J Sci 2(11):559–572. https://doi.org/10.1080/14786440109462720

Lundberg Scott M and Su-In Lee (2017) A unified approach to interpreting model predictions, Adv Neural Informat Process Syst 30

Ribeiro, Marco Tulio, Sameer Singh, and Carlos Guestrin (2016) Model- agnostic interpretability of machine learning. arXiv preprint arXiv:1606.05386

Khanna VV, Chadaga K, Sampathila N, Prabhu S, Bhandage V, Hegde GK (2023) A distinctive explainable machine learning framework for detection of polycystic ovary syndrome. Appl Syst Innov 6:2

Google Scholar

Broløs KR, Machado MV, Cave C, Kasak J, Stentoft-Hansen V, Batanero VG, Wilstrup C (2021) An approach to symbolic regression using feyn

Hubert B, Wojciech K, Piotr P, Jakub W, Przemyslaw B (2021) Dalex: responsible machine learning with interactive, explainability and fairness in python. J Mach Learn Res 22:1

MathSciNet Google Scholar

Dutta KK, Manohar P, Krishnappa I (2024) Seizure stage detection of epileptic seizure using convolutional neural networks. Int J Elect Comp Eng (IJECE) 14:2

Zarei and BM Asl (2021) Automatic seizure detection using orthogonal matching pursuit, discrete wavelet transform, and entropy based features of EEG signals, Comp Biol Med 131

KK Dutta, P Manohar, S Rajagopalan, F Naaz, and M Lakshminarayanan (2022) Eye state detection from electro-encephalography signals using machine learning techniques,” in 2022 IEEE 2nd Mysore Sub Section International Conference (MysuruCon), pp. 1–7

Farooq MS, Zulfiqar A, Riaz S (2023) Epileptic seizure detection using machine learning: taxonomy, opportunities, and challenges. Diagnostics 13:6

Dutta KK, Manohar P, Indira K (2023) Time and frequency domain pre-processing for epileptic seizure classification of epileptic EEG signals. J Intell Fuzzy Syst 45(5):8217–8226

Lahmiri S, Shmuel A (2018) Accurate classification of seizure and seizure- free intervals of intracranial EEG signals from epileptic patients. IEEE Trans Instrum Meas 68:791–796

Fasil OK, Rajesh R (2019) Time-domain exponential energy for epileptic EEG signal classification. Neurosci Lett 694:1–8

Siddiqui MK, Islam Z, Kabir MA (2018) A novel quick seizure detection and localization through brain data mining on ECoG dataset. Neural Comput Appl 31:5595–5608

Selvakumari RS, Mahalakshmi M, Prashalee P (2019) Patient-specific seizure detection method using hybrid classifier with optimized electrodes. J Med Syst 43:121

Rabcan J, Levashenko V, Zaitseva E, Kvassay M (2021) EEG signal classification based on fuzzy classifiers. IEEE Trans Ind Inform 18:757–766

Raghu S, Sriraam N, Vasudeva Rao S, Hegde AS, Kubben PL (2020) Automated detection of epileptic seizures using successive decomposition index and support vector machine classifier in long-term EEG. Neural Comput Appl 32:8965–8984

Omidvar M, Zahedi A, Bakhshi H (2021) EEG signal processing for epilepsy seizure detection using 5-level Db4 discrete wavelet transform, GA-based feature selection and ANN/SVM classifiers. J Ambient Intell Humaniz Comput 12:10395–10403

Pattnaik S, Rout N, Sabut S (2022) Machine learning approach for epileptic seizure detection using the tunable-Q wavelet transform based time– frequency features. Int J Inf Technol 14:3495–3505

Harpale V, Bairagi V (2021) An adaptive method for feature selection and extraction for classification of epileptic EEG signal in significant states. J King Saud Univ Comput Inf Sci 33:668–676

Amin HU, Yusoff MZ, Ahmad RF (2020) A novel approach based on wavelet analysis and arithmetic coding for automated detection and diagnosis of epileptic seizure in EEG signals using machine learning techniques. Biomed Signal Process Control 56:101707

Pattnaik S, Rout N, Sabut S (2022) Machine learning approach for epileptic seizure detection using the tunable-Q wavelet transform based time–frequency features. Int J Inf Tecnol 14:3495–3505

Basha NK, Surendiran B, Benzikar A et al (2024) Hybrid approach for the detection of epileptic seizure using electroencephalography input. Int J Inf Tecnol 16:569–575

Das P, Nanda S (2023) A novel multivariate approach for the detection of epileptic seizure using BCS-WELM. Int J Inf Tecnol 15:149–159

Rani TP, Chellam GH (2021) A novel peak signal feature segmentation process for epileptic seizure detection. Int J Inf Tecnol 13:423–431

Kumar PR, Shilpa B, Jha RK et al (2023) A novel end-to-end approach for epileptic seizure classification from scalp EEG data using deep learning technique. Int J Inf Tecnol 15:4223–4231

Rukhsar S (2022) Discrimination of multi-class EEG signal in phase space of variability for epileptic seizure detection using error correcting output code (ECOC). Int J Inf Tecnol 14:965–977

Download references

Open access funding provided by Manipal Academy of Higher Education, Manipal.

Author information

Shoibolina Kaushik, Diana Olivia and Zaid Khan authors contributed equally to this work.

Authors and Affiliations

Manipal Institute of Technology, Manipal, Karnataka, 576104, India

Shoibolina Kaushik & Zaid Khan

Department of Computer Science and Engineering, Manipal Institute of Technology, Manipal Academy of Higher Education, Manipal, Karnataka, India

Mamatha Balachandra

Department of Information and Communication Technology, Manipal Institute of Technology, Manipal Academy of Higher Education, Manipal, Karnataka, India

Diana Olivia

You can also search for this author in PubMed Google Scholar

Corresponding authors

Correspondence to Mamatha Balachandra or Diana Olivia .

Ethics declarations

Conflict of interest.

The authors declare that they have no competing interests that could potentially influence the content of this work.

Rights and permissions

Reprints and permissions

About this article

Kaushik, S., Balachandra, M., Olivia, D. et al. Unveiling the epilepsy enigma: an agile and optimal machine learning approach for detecting inter-ictal state from electroencephalogram signals. Int. j. inf. tecnol. (2024). https://doi.org/10.1007/s41870-024-02078-4

Download citation

Received : 10 April 2024

Accepted : 09 July 2024

Published : 23 August 2024

DOI : https://doi.org/10.1007/s41870-024-02078-4

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

EEG signals
Improved quality of life
Machine learning
Medical diagnosis
Find a journal
Publish with us
Track your research

Study assesses seizure risk from stimulating thalamus

In awake mice, researchers found that even low stimulation currents could sometimes still cause electrographic seizures

The idea of electrically stimulating a brain region called the central thalamus has gained traction among researchers and clinicians because it can help arouse subjects from unconscious states induced by traumatic brain injury or anesthesia, and can boost cognition and performance in awake animals. But the method, called CT-DBS, can have a side effect: seizures. A new study by researchers at MIT and Massachusetts General Hospital (MGH) who were testing the method in awake mice, quantifies the probability of seizures at different stimulation currents and cautions that they sometimes occurred even at low levels.

“Understanding production and prevalence of this type of seizure activity is important because brain stimulation-based therapies are becoming more widely used,” said co-senior author Emery N. Brown, Edward Hood Taplin Professor of Medical Engineering and Computational Neuroscience in The Picower Institute for Learning and Memory, the Institute for Medical Engineering and Science, the Department of Brain and Cognitive Sciences and the Center for Brains Minds and Machines (CBMM) at MIT.

In the brain, the seizures associated with CT-DBS occur as “electrographic seizures” which are bursts of voltage among neurons across a broad spectrum of frequencies. Behaviorally, they manifest as “absence seizures” in which the subject appears to take on a blank stare and freezes for about 10-20 seconds.

In their study, the researchers were hoping to determine a CT-DBS stimulation current— in a clinically relevant range of under 200 microamps—below which seizures could be reliably avoided.

Two panels show the emergence of seizures. Calm, narrow dark blue squiggles representing a tight range of low voltages at various currents explode into vibrant, wide squiggles with warmer colors as current and voltage increases at a broadening range of frequencies.

In search of that ideal current, they developed a protocol of starting brief bouts of CT-DBS at 1 microamp and then incrementally ramping the current up to 200 microamps until they found a threshold where an electrographic seizure occurred. Once they found that threshold, then they tested a longer bout of stimulation at the next lowest current level in hopes that an electrographic seizure wouldn’t occur. They did this for a variety of different stimulation frequencies. To their surprise, electrographic seizures still occurred 2.2 percent of the time during those longer stimulation trials (i.e. 22 times out of 996 tests) and in 10 out of 12 mice. At just 20 microamps, mice still experienced seizures in 3 out of 244 tests, a 1.2 percent rate.

“This is something that we needed to report because this was really surprising,” said co-lead author Francisco Flores, a research affiliate in The Picower Institute and CBMM, and an instructor in anesthesiology at MGH where Brown is also an anesthesiologist. Isabella Dalla Betta, a technical associate in The Picower Institute, co-led the study published in Brain Stimulation .

Stimulation frequency didn’t matter for seizure risk but the rate of electrographic seizures increased as the current level increased. For instance, it happened in 5 out of 190 tests at 50 microamps, and 2 out of 65 tests at 100 microamps. The researchers also found that when an electrographic seizure occurred, it did so more quickly at higher currents than at lower levels. Finally, they also saw that seizures happened more quickly if they stimulated the thalamus on both sides of the brain vs. just one side. Some mice exhibited behaviors similar to absence seizure, though others became hyperactive.

It is not clear why some mice experienced electrographic seizures at just 20 microamps while two mice did not experience the seizures even at 200. Flores speculated that there may be different brain states that change the predisposition to seizures amid stimulation of the thalamus. Notably, seizures are not typically observed in humans who receive CT-DBS while in a minimally conscious state after a traumatic brain injury or in animals who are under anesthesia. Flores said the next stage of the research would aim to discern what the relevant brain states may be.

In the meantime, the study authors wrote, “EEG should be closely monitored for electrographic seizures when performing CT-DBS, especially in awake subjects.”

The paper’s co-senior author is Matt Wilson, Sherman Fairchild Professor in The Picower Institute, CBMM, and the departments of Biology and Brain and Cognitive Sciences. In addition to Dalla Betta, Flores, Brown and Wilson, the study’s other authors are John Tauber, David Schreier, and Emily Stephen.

Support for the research came from The JPB Foundation, The Picower Institute for Learning and Memory, George J. Elbaum (MIT ‘59, SM ‘63, PhD ‘67), Mimi Jensen, Diane B. Greene (MIT, SM ‘78), Mendel Rosenblum, Bill Swanson, annual donors to the Anesthesia Initiative Fund; and the National Institutes of Health.

Method enables fast, accurate estimates of cardiovascular state to inform blood pressure management.

Cartoon image of a heart in the background with a major artery extending into the foreground. The artery is cut away at one point to show a flow of red blood cells within.

Study reveals how an anesthesia drug induces unconsciousness

Two vials of propofol, a milky liquid, rest on a hospital shelf next to a small yelllow bin

Consciousness

With programmable pixels, novel sensor improves imaging of neural activity

A composite of images features the whole imaging chip which includes a black rectangle in the top third labeled as the pixel area, a more orange below labeled as the amplifiers and analog digital converters and an outer area that contains the digital controls and readout. The whole chip is about 5 mm wide by 4 mm tall. An inset shows individual pixels, which are mostly a black photodiode with a thin strip on the left of smaller circuitry. A circuit schematic shows an arrangement of six pixel transistors.

COMMENTS

Current Classification of Seizures and Epilepsies: Scope, Limitations and Recommendations for Future Action
The classification of seizures and epilepsies by the International League Against Epilepsy (ILAE), 2017 is the most recent classification model which aimed to simplify terminologies that patients and their caregivers can easily understand, identify seizures that have both focal and generalized onset and incorporate missing seizures.
New Trends and Most Promising Therapeutic Strategies for Epilepsy
Introduction. Epilepsy is the enduring predisposition of the brain to generate seizures, a condition that carries neurobiological, cognitive, psychological, and social consequences ().Over 50 million people worldwide are affected by epilepsy and its causes remain partially elusive, leaving physicians, and patients an unclear insight into the etiology of the disease and the best treatment ...
Seizures and Epilepsy: An Overview for Neuroscientists
"Epileptic seizure" is used to distinguish a seizure caused by abnormal neuronal firing from a nonepileptic event, such as a psychogenic seizure. "Epilepsy" is the condition of recurrent, unprovoked seizures. Epilepsy has numerous causes, each reflecting underlying brain dysfunction (Shorvon et al. 2011). A seizure provoked by a ...
Epilepsy: A Clinical Overview
The diagnosis and treatment of seizures and epilepsy is a common task of the physician. Approximately 1 in 10 people will have a seizure during their lifetime. Epilepsy is the tendency to have unprovoked seizures. Epilepsy is the fourth most common neurological disorder and affects 1 in 26 people in the United States and 65 million people worldwide. Evaluation of a patient presenting with a ...
Epilepsy research in 2022: clinical advances
In 2022, epilepsy research has made advances across a range of clinically important areas, from self-management, genetics, imaging, and surgical planning to understanding febrile seizures and coma-related periodic patterns. Most notably, in May 2022, the World Health Assembly adopted the Intersectoral Global Action Plan on Epilepsy and Other Neurological Disorders, which aims to address gaps ...
Epilepsy Research
Epilepsy Research provides for publication of high quality articles in both basic and clinical epilepsy research, with a special emphasis on translational research that ultimately relates to epilepsy as a human condition. The journal is intended to provide a forum for reporting the best and most …. View full aims & scope.
Epilepsy
Epilepsy articles from across Nature Portfolio. Epilepsy refers to a group of neurological disorders of varying aetiology, characterized by recurrent brain dysfunction that result from sudden ...
Major advances in epilepsy research in 2023
With great anticipation, 2023 has seen many important advances in epilepsy research. Noteworthy progress has been achieved in understanding the intricate mechanisms of epilepsy, accompanied by important strides in developing new therapies. Historically, the efficacy of most second-generation and third-generation antiseizure medications in ...
Epilepsy-Definition, Classification, Pathophysiology, and Epidemiology
Seizures affect the lives of 10% of the global population and result in epilepsy in 1 to 2% of people around the world. Current knowledge about etiology, diagnosis, and treatments for epilepsy is constantly evolving. As more is learned, appropriate and updated definitions and classification systems …
Initial Management of Seizure in Adults
The Medical Research Council Multicentre Trial for Early Epilepsy and Single Seizures 24 showed that the risk of seizure recurrence was lower in the first 2 years after the first seizure among ...
The pharmacological treatment of epilepsy: recent advances and future
The pharmacological armamentarium against epilepsy has expanded considerably over the last three decades, and currently includes over 30 different antiseizure medications. Despite this large armamentarium, about one third of people with epilepsy fail to achieve sustained seizure freedom with currently available medications. This sobering fact, however, is mitigated by evidence that clinical ...
Epilepsia
Epilepsia. Epilepsia is the leading, authoritative journal for innovative clinical and basic science research for all aspects of epilepsy and seizures. In addition, Epilepsia publishes critical reviews, opinion pieces, and guidelines that foster understanding and aim to improve the diagnosis and treatment of people with seizures and epilepsy.
Seizures, semiology, and syndromes: A narrative review
Semiology, a system of objective and subjective seizure signs, guides us towards the classification of seizure types and epilepsy syndromes. •. Through genotype-phenotype correlations and automated reasoning algorithms, information on seizures and epilepsy syndromes continues to be of critical importance to ongoing research.
Journal of Epilepsy Research
ORIGINAL ARTICLE 2024 June 30 Medication Reconciliation Errors on Discharge for Epilepsy Monitoring Unit Patients: Background and Purpose: Medication errors are common in the inpatient setting. Epilepsy patients who miss doses of their antiseizure medications are at risk for breakthrough seizures and subsequent complications.
Revisiting the role of neurotransmitters in epilepsy: An ...
Epileptic seizures maydevelop due to a relative imbalance of excitatory and inhibitory neurotransmitters. Expressional alterations of receptors and ion channelsactivated by neurotransmitters can lead to epilepsy pathogenesis. ... The PubMed database was searched for related research articles. Key findings: Glutamate and gamma-aminobutyric acid ...
Epilepsy Research News -- ScienceDaily
Functional, Nonepileptic Seizures Show Structural Abnormalities in Brain Scans, Study Shows. Oct. 25, 2022 — Functional seizures not caused by epilepsy are associated with structural changes in ...
Epilepsy Research
Read the latest articles of Epilepsy Research at ScienceDirect.com, Elsevier's leading platform of peer-reviewed scholarly literature. ... The utility of Multicentre Epilepsy Lesion Detection (MELD) algorithm in identifying epileptic activity and predicting seizure freedom in MRI lesion-negative paediatric patients. Aimee Goel, Stefano Seri ...
Epileptic seizures
Epileptic seizures are abnormal jerky or trembling movements in the body due to abnormal neuronal activity and can result in damage to the brain or other parts of the body. Even a single seizure can cause changes in neural development and can lead to behavioural and cognitive changes. ... Epilepsy Research. 2010; 89 (2-3):310-318. [Google ...
Family reflections: a candid view from caretakers of patients with
Childhood epilepsy is one of the most common pediatric neurologic diseases worldwide, affecting nearly 20 million children and their families, and is associated with increased morbidity and ...
Alterations in serum metabolomics during the first seizure and ...
Epilepsy is one of the most common brain diseases, which is characterized by repetitive, episodic, and transient central nervous system dysfunction caused by excessive discharge of brain neurons 1 ...
Pathophysiological mechanisms underlying the development of focal
Epilepsy is characterized by recurrent episodes of paroxysmal brain dysfunction caused by sudden, synchronous, and excessive neuronal discharge. Focal cortical dysplasia (FCD) is a form of structural lesion with different sizes, locations, and histopathological manifestations [1]. It is characterized by abnormal non-neoplastic cell proliferation in the cerebral cortex, confined to a region of ...
Pathophysiological mechanisms underlying the development of focal
Drug-resistant epilepsy caused by FCD is a major issue in clinical practice. This article reviews the current clinical presentations of FCD, factors related to the mTOR pathway, and animal models of FCD. The objective was to provide a theoretical basis for research on FCD [54]. 2. FCD clinical presentation
Adverse Events of Cannabidiol Use in Patients With Epilepsy
Epilepsy is one of the most common neurologic disorders globally, with a lifetime point prevalence of 7.6 per 1000 population and an annual incidence of 67 per 100 000 population. 1 Although most cases can be treated or go into remission with age, in approximately one-third of cases the seizures continue despite pharmacotherapy, surgical, or ...
Management of Psychogenic Nonepileptic Seizures
In their review of the epidemiology of psychogenic nonepileptic seizures, Asadi-Pooya and Sperling reported that psychogenic seizures are relatively common, since they are reported to be experienced by 5%-10% of outpatients in epilepsy clinics and 20%-40% of inpatients in epilepsy monitoring units.In three studies reviewed by Asadi-Pooya and Sperling, the incidence of psychogenic ...
Psychogenic nonepileptic seizure: An empathetic, practical approach
Psychogenic nonepileptic seizure (PNES) is often misdiagnosed as epilepsy, leading to unnecessary treatments and procedures, as well as failure to engage patients in needed mental health care. To establish an accurate diagnosis, video electroencephalography (EEG) in the context of and simultaneous with a comprehensive neurologic and psychosocial evaluation is recommended for any patient with ...
Relationship between anti-seizure medication and behaviors that
Epilepsy is a chronic neurological disorder with a broad etiology and a heterogeneous spectrum of clinical manifestations. ... b University of South Wales, Wales, UK;c Cornwall Intellectual Disability Research (CIDER), Peninsula Schools of Medicine and Dentistry, University of ... including anti-seizure medications (ASM), in people with ...
Recent advances in epilepsy
Abstract. This paper reviews advances in epilepsy in recent years with an emphasis on therapeutics and underlying mechanisms, including status epilepticus, drug and surgical treatments. Lessons from rarer epilepsies regarding the relationship between epilepsy type, mechanisms and choice of antiepileptic drugs (AED) are explored and data ...
High-Field 7T MRI in a drug-resistant paediatric epilepsy cohort: image
Background and Objectives: Epileptogenic lesions in focal epilepsy can be subtle or undetected on conventional brain MRI. Ultra-high field (7T) MRI offers higher spatial resolution, contrast and signal-to-noise ratio compared to conventional imaging systems and has shown promise in the pre-surgical evaluation of adult focal epilepsy. However, the utility of ultra-high field MRI in paediatric ...
Unveiling the epilepsy enigma: an agile and optimal machine ...
Epilepsy is a chronic neurological disorder characterized by the occurrence of paroxysmal recurrent seizures, which are caused by abnormal electrical activity in the brain. Seizures vary widely in their presentation, depending on the specific region of the brain involved and the extent of the abnormal electrical discharges. The disease can affect cognitive function posing a serious threat to ...
Study assesses seizure risk from stimulating thalamus
They did this for a variety of different stimulation frequencies. To their surprise, electrographic seizures still occurred 2.2 percent of the time during those longer stimulation trials (i.e. 22 times out of 996 tests) and in 10 out of 12 mice. At just 20 microamps, mice still experienced seizures in 3 out of 244 tests, a 1.2 percent rate.

The pharmacological treatment of epilepsy: recent advances and future perspectives

Advances in characterizing the comparative effectiveness and safety of ASMs

Introduction of novel ASMs

Introduction of novel formulations

Future perspectives

Increased application of technological tools to epilepsy management

Precision therapies

Biomarker-guided therapies

Novel drugs and the search for disease-modifying therapies

Conclusions

Availability of data and materials

Abbreviations

Acknowledgments

Author information

Contributions

Corresponding author

Ethics declarations

Competing interests

Additional information

Rights and permissions

About this article

Share this article

Acta Epileptologica

Save citation to file

Add to My Bibliography

Revisiting the role of neurotransmitters in epilepsy: An updated review

Similar articles

Publication types

Related information

LinkOut - more resources

Other Literature Sources

Alterations in serum metabolomics during the first seizure and after effective control of epilepsy

Similar content being viewed by others

NMR-based metabolomics in pediatric drug resistant epilepsy – preliminary results

Identification of cerebrospinal fluid and serum metabolomic biomarkers in first episode psychosis patients

Targeted metabolomics reveals aberrant profiles of serum bile acids in patients with schizophrenia

The research protocol and flow chart

Basic information and clinical characteristics of participants

Characterization of serum metabolomics of EPs and establishment of diagnostic model

Association analysis of oral microecology, serum metabolome, and laboratory indicators in Eps and HCs

Characterization of the serum metabolites in EPRs and HCs

Correlation analysis of the oral microecology, metabolic spectrum, and clinical indicators in EPRs and HCs

Characterization of serum metabolites before and after seizure control

Correlation analysis of the oral microecology, metabolic spectrum, and clinical indicators before and after seizure control

Research program

Serum sample preparation

LC–MS experiment

Chromatogram of total ions

Identification of metabolites

Multivariate data analysis

Advanced analysis

Construction of diagnostic models

Statistical analysis

Limitations of the study

Ethics approval and consent to participate

Patient and public involvement

Data availability

Acknowledgements

Author information

Authors and Affiliations

Contributions

Corresponding authors

Ethics declarations

Additional information

Supplementary Information

About this article

Share this article

Quick links

See More About

Get the latest research based on your areas of interest.

Manage citations:

Adverse Events of Cannabidiol Use in Patients With Epilepsy : A Systematic Review and Meta-analysis

Management of Psychogenic Nonepileptic Seizures

Conclusions

Export Citations

View options

Not a subscriber?

Share article link

NEXT ARTICLE

Create a new account