Insights into epileptogenesis from post-traumatic epilepsy

Abstract

Post-traumatic epilepsy (PTE) accounts for 5% of all epilepsies. The incidence of PTE after traumatic brain injury (TBI) depends on the severity of injury, approaching one in three in groups with the most severe injuries. The repeated seizures that characterize PTE impair neurological recovery and increase the risk of poor outcomes after TBI. Given this high risk of recurrent seizures and the relatively short latency period for their development after injury, PTE serves as a model disease to understand human epileptogenesis and trial novel anti-epileptogenic therapies. Epileptogenesis is the process whereby previously normal brain tissue becomes prone to recurrent abnormal electrical activity, ultimately resulting in seizures. In this Review, we describe the clinical course of PTE and highlight promising research into epileptogenesis and treatment using animal models of PTE. Clinical, imaging, EEG and fluid biomarkers are being developed to aid the identification of patients at high risk of PTE who might benefit from anti-epileptogenic therapies. Studies in preclinical models of PTE have identified tractable pathways and novel therapeutic strategies that can potentially prevent epilepsy, which remain to be validated in humans. In addition to improving outcomes after TBI, advances in PTE research are likely to provide therapeutic insights that are relevant to all epilepsies.

Introduction

Traumatic injuries account for more lost years and productivity than cancer or cardiovascular disease¹. Affecting 55 million people worldwide, traumatic brain injury (TBI) is a leading cause of death and disability^2,3. Post-traumatic epilepsy (PTE) is a major driver of the neurological, emotional and occupational disability associated with TBI. Seizures cause secondary injury after TBI, often exacerbating neurodegeneration, inflammation and blood–brain barrier (BBB) dysfunction^4,5. PTE impairs neurological recovery after TBI and is independently associated with poor functional outcomes^6,7. PTE affects up to one-third of individuals with severe TBI and accounts for 20% of structural epilepsies overall, so effective treatment of this condition provides an opportunity to reduce the burden of disability after TBI^8,9.

PTE develops through epileptogenesis^10–12 — a chronic process whereby normally functioning brain tissue becomes prone to recurrent abnormal electrical activity that results in seizures. This continuous and prolonged process is traditionally thought to occur in three, often overlapping, clinical phases: first, an initial insult such as TBI; second, a latency period lasting months to years during which epileptic circuits develop; and third, the symptomatic phase with spontaneous, unprovoked seizures defining epilepsy¹³. Once epilepsy develops, epileptogenesis continues as seizures reinforce and expand epileptic circuits, potentially developing into drug-resistant epilepsy. Despite substantial progress in the development of over 30 anti-seizure medications (ASMs) over the past few decades, we still lack the ability to modify or alter the process of epileptogenesis in humans^12,14,15.

The development of disease-modifying anti-epileptogenic medications (AEMs) is recognized as a public health priority by many national and international organizations, including the WHO, the NIH and the European Commission^16–18. PTE is an excellent model disease to elucidate the mechanisms of epileptogenesis and trial AEMs, as the initial trigger (TBI) is common and clearly defined, and epileptogenesis typically occurs over a relatively short latency period. Epileptogenesis continues over the disease course, as patients with PTE have higher rates of drug resistance than do individuals with other epilepsy types^19,20. Multi-institutional consortia are developing clinical and preclinical trial networks to identify targeted therapies to better treat PTE^21–26.

In this Review, we first provide a clinical overview of PTE, describing its clinical characteristics, course and current treatment strategies. We then review epileptogenic mechanisms in PTE and tractable targets for AEMs. A major challenge in navigating the PTE research field is the range of different definitions for common terms. We define PTE as a single late (more than 7 days post-injury) post-traumatic seizure^6,11 (PTS). The International League Against Epilepsy (ILAE) defines epilepsy as two seizures at least 24 h apart or a single seizure with a high risk of recurrence²⁷ (more than 60% at 10 years). Owing to the high risk of seizure recurrence after TBI, a single late PTS (LPTS) meets the ILAE definition of PTE^11,28–30.

Epidemiology of PTE

Despite considerable efforts, the incidence of PTE has been difficult to define owing to inadequacies in seizure reporting and long-term follow-up and variations in study approaches. The available data indicate that PTE is common, with a prevalence of 50% among war veterans with penetrating TBI and around 33% among civilians with severe TBI²⁰. Worldwide, PTE accounts for 5% of the 70 million cases of epilepsy worldwide and 20% of symptomatic epilepsies with a structural aetiology^8,31,32. In a seminal study, the Rochester Epidemiology Project at the Mayo Clinic retrospectively studied PTE rates in patients with TBI from the 1930s to the 1980s¹⁰. This study established two guiding principles for PTE epidemiology: first, the rate of PTE rises considerably with increasing TBI severity; and second, PTE can occur years to decades after injury.

In the light of these findings, subsequent studies used prospective databases to better define the rate of PTE after TBI (Fig. 1). When the Glasgow Coma Scale (GCS) was used to stratify injury severity, the rates of PTE after 1–2 years of follow-up were 12–26% for severe TBI^7,29,33,34, 5–24% for moderate TBI^7,33,34 and 1–8% for mild TBI^7,34. Studies using the Abbreviated Injury Scale found similar results, with PTE rates ranging from 8% to 21% in people with moderate-to-severe TBI^35,36. The main strength of these studies was the ability to clearly define ranges for the risk of PTE up to 2 years after injury. The prohibitive cost of prospectively monitoring large cohorts of patients for years after their injuries limits our understanding of how PTE evolves over long time periods. Similarly, these prospective studies are limited by loss to follow-up, with some studies only having seizure data available for 15–50% of patients^7,36. The definition of seizures also varied, with many studies using non-validated epilepsy questionnaires^7,36.

Fig. 1 |. Prevalence of post-traumatic epilepsy after moderate or severe traumatic brain injury.

The graph shows the prevalence of post-traumatic epilepsy (PTE) over time after traumatic brain injury (TBI) for selected representative cohorts. Burke et al.⁷, Englander et al.³⁴, Temkin et al.⁸⁵, Ritter et al.²⁹ and Tubi et al.⁴⁷ prospectively recorded seizures using physician visits, patient forms or phone calls. Pease et al.¹¹ recorded seizures through retrospective chart review of a prospectively maintained database, whereas DeGrauw et al.³⁹ used insurance claims data. The data indicate that the PTE onset occurs during the first year post-TBI in most cases, although the risk of developing PTE persists for at least 5 years.

Another approach to tracking PTE incidence involved the use of large population-based electronic medical record systems^10,35,37,38 or claims databases³⁹. These studies had several important methodological differences from the prospective studies described earlier. First, many reported relative rather than absolute risk of developing PTE compared with matched controls^37,38. Second, none of these studies used the GCS to stratify injury severity, instead relying on various inconsistent definitions. In a seminal study of PTE risk⁴⁰, Annengers et al. found a hazard ratio of 17 for severe TBI (defined by the authors as brain contusion or loss of consciousness for >24 h), 2.9 for moderate TBI (defined as skull fracture or loss of consciousness or amnesia for >30 min) and 1.5 for mild TBI (defined as loss of consciousness or amnesia for <30 min). Finally, most of the studies used diagnostic codes or claims databases rather than patient-reported data to identify seizures^37,38,41. Cumulatively, these studies showed that the risk of PTE increased with TBI severity, with intracerebral contusions conferring the greatest risk. The rates of PTE after 1 year tended to be lower than that in prospective studies: typically <5% overall and <10% even in people with severe TBI^35,38,39,41. This discrepancy might reflect the difficulties of using medical coding or billing to identify seizures^42,43, and also the likelihood that PTE is underdiagnosed and undertreated in community settings.

Mild TBI with a negative CT head scan is 10 times more common than moderate or severe TBI, affecting nearly one million people in North America yearly⁴⁴, and is associated with the lowest rates of PTE post-injury. Although some older studies found that rates of PTE after this type of TBI were comparable to the annual incidence of epilepsy in the general population^10,40 (60 in 10,000 individuals), more recent studies have found an increased risk, albeit small^37,38,45. Repeated mild TBI also seems to increase the risk of epilepsy⁴⁶.

Several studies, mainly retrospective, have determined that PTE can occur up to four decades after the inciting TBI^{10,11,20,29,39,41,47}. A decade after TBI, the cumulative incidence of PTE in modern cohorts was 32–53% in people with more severe injuries^11,47,48. The increased risk continues across the spectrum of TBI severity, with patients who sustained a concussion having double the relative risk at a decade post-injury compared with the general population^10,39,41. Seizures that occur years to decades after injury seem to be similar to those occurring within the first year post-injury in terms of semiology and rates of treatment resistance²⁰. Despite the long tail of PTE risk, PTE onset typically occurs early after TBI, often within the first 6–12 months^6,10,34,41. This knowledge can inform the timing of PTE screening, with benefits not only for clinical management but also for clinical trial design.

A hallmark of PTE is its relative severity, including high seizure frequency and recurrence rates. The recurrence rates approach 80% in the first few years after index seizure^11,28–30, and almost half of the patients with PTE have five or more repeat seizures³⁰. The rates of medication-resistant PTE range from 23% to 53%, compared with 30–40% for non-traumatic epilepsies^11,17,20,30.

PTE is understudied in World Bank-designated low-and-middle-income countries, with many groups reporting incidences of 10% or less even in people with more severe injuries^28,49–52. This phenomenon probably reflects a combination of the difficulties of running expensive prospective studies in regions with limited resources, undertreatment and lack of surveillance for PTE. Nearly 80% of patients with epilepsy in low-income countries do not have access to treatment⁵³. The high rate (15%) of PTE in epilepsy clinics in low-income countries, which is similar to the prevalence of symptomatic epilepsy in high-income countries, suggests undertreatment rather than reduced prevalence.

Risk factors for PTE

The main driver of PTE risk is injury severity, typically measured by the GCS and proxy measures of brain damage. However, the GCS is an imperfect measure of injury severity; for example, it has a loose correlation with mortality⁵⁴, and prognostic models incorporating the GCS account for less than one-third of the variation in outcomes^55,56. In multivariate prediction models for TBI, the GCS is frequently replaced by other markers of injury severity, including duration of post-traumatic amnesia^57,58, midline shift⁵⁹ and neurosurgical procedures^11,58,60. Decompressive craniectomy, whereby nearly one-third of the calvarium of the skull is removed to accommodate swelling, is associated with an odds ratio of more than 4 for PTE in multivariate models, although this factor probably correlates with injury severity rather than being causative^11,58.

Age correlates inversely with PTE risk, with younger adults at greater risk than older adults^11,29. This finding could reflect the fact that younger patients are more likely to survive severe injuries and go on to develop PTE or, alternatively, that they have greater CNS plasticity, which increases susceptibility to the formation of epileptic circuits. Age is a major predictor of survival in most clinical prognostic models^61–63.

Most markers of injury severity correlate with underlying brain damage, which can be causative for PTE but is challenging to quantify^64,65. For example, healing and gliosis after focal cortical injury could induce epileptogenesis by promoting the development of abnormal electric circuits in the cortex⁶⁶. These circuits are reinforced into a network, thereby maintaining aberrant hypersynchronized electrical activity and causing seizures. One study showed that veterans with penetrating brain injuries, which often result in focal cortical injury, had more than a 50% lifetime risk of developing PTE⁴⁸. Individuals with depressed skull fractures, who often have underlying cortical damage, also have a clinically increased risk of PTE^59,67,68.

Deeper focal injuries also pose a risk of PTE⁶⁴. Subcortical contusions, which also lead to abnormal healing and gliosis, are associated with an increased PTE risk^33,69. Greater numbers and sizes of brain lesions, as well as increased surrounding oedema, increase PTE risk in a volume-dependent manner^58,64,70. Incomplete gliosis surrounding focal lesions independently increases PTE risk, as healing occurs in a haphazard, abnormal pattern⁷¹. Studies from specialized epilepsy surgery clinics have corroborated the relationship between focal injuries and PTE. Patients in these clinics, who typically have long-standing PTE, predominantly have focal epilepsy, with almost one-half having a cortical focus⁷². As in the case of other focal epilepsies, temporal lobe seizures are the most common, followed by frontal lobe seizures; parietal and occipital foci are relatively uncommon⁷². This finding is probably attributable not only to the high prevalence of temporal lobe damage in TBI but also to the underlying differential epileptogenic potential of various brain regions. Large mass-occupying lesions that deform the brain parenchyma, such as subdural haematomas, are associated with underlying injury and greatly increased risks of PTE, with or without surgery^11,59,60,73. By contrast, epidural haematomas, typically from skull fractures and laceration of the meningeal vessels, do not seem to increase PTE risk^11,59.

Diffuse or extra-axial injuries also contribute to PTE. Diffuse axonal injury results from shear stress along white matter tracts. Damage to these tracts can persist for years and is prominent in patients with cognitive decline after TBI⁷². These diffuse injuries can modify white matter circuitry through Wallerian degeneration and sprouting of new aberrant, potentially epileptogenic, connections^64,74. In addition, diffuse injuries increase PTE risk by damaging key epileptogenic brain regions such as the hippocampus⁷⁵ or thalamus⁷⁶, which modulate seizure suppression through network changes. Measuring diffuse injuries in the acute phase after TBI can be challenging.

Secondary brain injuries further exacerbate PTE risk. In the acute phase, elevations in intracranial pressure can cause widespread damage to white matter tracts, as well as neuroinflammation and neurodegeneration^77,78. In the longer term, intracranial infections can also increase the risk of PTE^11,79. Compared with a single TBI, repeated TBI months to years after the initial injury nearly doubles the PTE risk⁴⁶.

The risk factors that affect PTE latency — that is, the time from injury to first seizure onset — are largely unknown. As we highlighted earlier, epidemiological studies have demonstrated that PTE onset can occur decades after injury^29,33,39, although it typically manifests within the first few months up to a year after injury^10,11,34 (Fig. 1). Only one study has evaluated risk factors for reduced PTE latency and found that age, injury severity and residual disability were associated with decreased time to PTE onset. Risk of seizure recurrence, although known to be high^11,29,30, has not been studied in detail, but might be reduced by cerebrospinal fluid shunting¹¹.

Subpopulations of people with PTE, such as those stratified by gender and socioeconomic status, have not been adequately studied. Males seem to have a higher risk of PTE and seizure recurrence than do females, although one group found that this phenomenon only applied to individuals with severe TBI^46,59,80. Groups with disadvantaged social support networks, such as prisoners or those with alcohol misuse, are also at increased risk of PTE^46,59,80. Race does not seem to influence PTE risk, although this has not been studied thoroughly²⁹.

Early PTSs and PTE

PTSs are divided into early PTSs (EPTSs), which manifest within 7 days of trauma, and LPTSs, which manifest beyond 7 days¹¹. EPTSs are considered to be provoked seizures, whereas LPTSs are unprovoked and are synonymous with PTE. EPTS rates vary depending on the injury severity, ranging from 0% to 4% for mild TBI and up to 15% without prophylaxis after severe TBI^81–85. EPTSs are associated with increased length of hospital stay and with non-home discharge and mortality³⁶.

Some groups have further subdivided EPTSs to include an immediate seizure subgroup⁸⁶. Although definitions vary, immediate seizures typically occur in the pre-hospital setting or before resuscitation. Early trauma research indicated that immediate PTSs did not confer an increased risk of PTE, as they were typically acutely provoked by injury rather than by the development of epileptogenic circuits in the brain^29,40,87. Differentiating between seizures and pathological posturing responses can also be challenging in the pre-hospital setting. Some researchers have classified immediate EPTSs as any seizure within the first 24 h after TBI, often for the sake of convenience in large coding-based databases, in which it can be difficult to distinguish between seizure onset in pre-resuscitation and post-resuscitation settings^29,88. Considering that most EPTSs occur during the first day^85,89, this approach might inappropriately increase the risk attributed to immediate seizures and explain why one group found that immediate EPTSs increased PTE risk, contrary to other groups²⁹.

Randomized controlled trials (RCTs), meta-analyses and guidelines indicate that ASM prophylaxis protects against EPTSs in people with severe TBI^85,90–92, providing an absolute risk reduction of nearly 11% (ref. 85). Clinical trials routinely measure ASM levels in the peripheral blood and adjust the dose as appropriate, although this is rarely done in clinical practice, which could affect real-world clinical outcomes^93,94. All types of ASMs seem to be equally effective, although sodium valproate use increased mortality in one study⁹⁵; most practitioners use levetiracetam or phenytoin^92,96,97. Owing to the comatose nature of patients with severe TBI and their risk of electrographic seizures, guidelines recommend continuous EEG^36,94,98,99.

The robust effectiveness of ASM prophylaxis in severe TBI has not been reproduced in people with mild or moderate TBI^{83,88,100–102}, possibly because many studies are underpowered to find an effect owing to lower EPTS rates and/or smaller absolute risk reduction in mild and moderate TBI. The benefits of ASM prophylaxis in mild and moderate TBI should be weighed against the risks of inappropriate continuation of ASM prophylaxis^82,103, including infection¹⁰⁴, agitation¹⁰⁵, hyponatraemia¹⁰⁶ and poor functional outcomes¹⁰³.

EPTSs could provide important insights into epileptogenesis. Although the data are inconsistent, EPTS seems to confer an increased risk of PTE^{11,28,34,35,39,47,52,107}, possibly through the early development of abnormal epileptogenic circuits. Patients with EPTSs have a shorter time to onset of PTE than those with LPTSs, although this association might be spurious as both EPTS and PTE latency are influenced by injury severity⁸⁰. In contrast to people with EPTSs, ASM prophylaxis does not reduce the long-term risk of PTE^85,92, probably reflecting the failure of current ASMs to modify or ameliorate epileptogenesis¹⁰⁸. ASMs are seizure-management tools, and modern ASMs have similar efficacy to older generations of ASMs^17,108. Although we cannot reliably modify epileptogenesis with ASMs at present, management of EPTSs might have a role in preventing PTE development — a possibility that should be explored in future studies.

PTE and long-term TBI outcomes

Whether PTE results in worsened disability after TBI, independently of injury severity, remains unproven, possibly owing to the large confounding effect of injury severity on both PTE risk and long-term outcomes. PTE is associated with worse outcomes when global measures, such as the Glasgow Outcome Scale (GOS) or the Extended GOS (GOSE)^6,7,47, functional independence measures^103,109,110, employment¹⁹, mental health^111–113 and cognitive decline¹¹⁴, are considered. However, it is difficult to fully account for the effects of PTE independently of injury severity, as markers of injury severity, such as GCS scores and pupil reactivity, have similar associations. Nonetheless, there are several mechanistic reasons why seizures might worsen neurological outcomes. PTE acts synergistically with injury severity, causing a ‘second hit’. Seizures worsen neural inflammation, which can promote neuronal cell death and remodelling of cognitive circuits, thereby limiting recovery from brain injury and exacerbating functional decline^115,116. In addition, seizures can promote BBB breakdown, which can result in neurodegeneration.

Identifying the relative contribution of PTE to poor outcomes after TBI is challenging. Two studies controlled for injury severity using propensity matching and regression, and they found that PTE was an independent risk factor for poor functional outcomes based on GOS scores^6,7. These approaches are limited, however, by difficulties in defining injury severity in TBI. Both large-scale, regression-based clinical prognostic models and advanced deep learning approaches account for less than one-third of the variation in outcomes^61–63,117. Other markers of injury severity, such as delayed subcortical atrophy, thalamic damage and penetrating injury, have not been measured in PTE studies, but might correlate better with long-term outcomes than do more traditional measures such as neurosurgical interventions, and also correlate closely with PTE^76,118,119.

Another limitation of the existing studies is that they rarely assessed whether PTE was well controlled or whether it was medically refractory. Epilepsy that is well controlled with medical therapy is not usually associated with long-term functional decline, whereas medically refractory epilepsy, particularly with temporal lobe onset, has been implicated in progressive neurodegeneration and dementia.

Nevertheless, the temporal relationship between PTE and poor outcomes suggests a causal link. Although early functional outcomes after TBI are similar between patients with and without PTE, the outcomes diverge over time^6,111,120. Nearly 15% of patients who have achieved favourable outcomes (GOSE score 5–8) at the time of PTE diagnosis will subsequently decline to unfavourable outcomes¹²¹ (GOSE score 2–4). Patients with longer latencies to PTE diagnosis tend to have better outcomes than those with shorter latencies, which might reflect reduced impairment in early recovery¹²¹. These temporal trends suggest but do not prove that PTE is an adverse independent prognosticator and second hit after TBI.

PTE is likely to increase long-term mortality risk, although this relationship is similarly confounded by correlations with injury severity^122–124. Among survivors of the index injury, PTE does not seem to increase mortality during the early post-traumatic period^6,122 (1–2 years post-injury). As with functional outcomes, the effects of PTE on mortality seem to be compounded over the long term^122–124 (years to decades after injury). One study that evaluated death records found that PTE was a leading cause of death in young adults who survived the acute TBI, highlighting a possible causal link between PTE and mortality¹²³.

TBI and epilepsy are notable risk factors for mental health and behavioural problems and cognitive decline. Nearly 60% of survivors of TBI experience psychiatric illness in the year after their injury, and half of patients with epilepsy experience a major depressive episode in their lifetime^125,126. PTE and mental health are likely to have a bidirectional relationship: both are associated with damage of emotion-regulating brain networks in the prefrontal and medial brain regions and alter the hypothalamic–pituitary–adrenal axis, thereby affecting mood and behavioural regulation. TBI survivors with PTE are at increased risk of new psychiatric or behavioural conditions, even after accounting for mental health history¹¹¹. Cognitive networks seem to sustain similar injury patterns, as early seizures and abnormal EEG activity were associated with worse cognitive outcomes in a cohort with similar global outcome measures^127,128. In the long term, PTE exacerbates the cognitive decline that is prevalent after TBI¹¹⁴.

Current treatment approaches and outcomes

PTE treatment is highly variable, and the outcome measures are poorly defined, with no formal guidelines for treatment or screening from major societies. Although the onset of PTE mostly occurs during the first year post-TBI, few institutions regularly screen for PTE as part of the routine follow-up. Of specific concern, despite high rates of pharmacoresistance, people with PTE have low rates of referral to specialized epilepsy centres, and surgery is probably underutilized^19,108: although it constitutes 20% of epilepsies with structural aetiology, PTE only accounts for 5% of patients in surgical epilepsy clinics^72,129.

In clinical practice, PTE is treated similarly to other symptomatic epilepsies, with ASMs as the mainstay of treatment. Literature on the effectiveness of certain ASMs or combinations specifically for PTE is scarce, and owing to historical trials that demonstrated the effectiveness of phenytoin for EPTS prophylaxis, patients with PTE are disproportionately prescribed older-generation ASMs, often by neurotrauma practitioners rather than epilepsy specialists^19,20. Newer-generation seizure medications, in particular levetiracetam, might have similar efficacy¹³⁰ and be better choices given their limited interaction profile and metabolic toxicity and favourable cognitive adverse effect profile^131,132. As in other focal epilepsies, pharmacoresistance, defined as the failure of at least two appropriate ASMs to adequately control seizures, represents a major treatment challenge in PTE¹³³. Younger age of PTE onset, focal or mixed onset seizure type, presence of status epilepticus and interictal epileptiform discharges are all predictors of pharmacoresistance⁸⁰.

Despite the high prevalence of pharmacoresistance, epilepsy surgery evaluations rarely seem to be conducted in PTE populations^72,134. This underutilization has been attributed to various factors, including concern about multifocality, distinct pathophysiological mechanisms and anatomical heterogeneity^135–137. A meta-analysis found higher rates of unfavourable post-surgical outcomes (Engel Epilepsy Surgery Outcome Scale Class II–IV) in temporal lobe PTE than in non-traumatic epilepsy¹³⁸. Although the result was statistically significant, the absolute reduction in Engel Class I (seizure freedom) rates was 5%, which was not a clinically meaningful difference as the overall Engel Class I rate (70.1%) remained comparable to historical trials¹³⁹. This result extended to extratemporal lobe PTE, with similarly excellent outcomes reported in patients with focal encephalomalacia on MRI who underwent intracranial EEG evaluations¹³⁶.

Neuromodulation might also have a role in the treatment of pharmacoresistant PTE. In one study, significant seizure reduction was observed in around two-thirds of patients with this condition who were treated with vagus nerve stimulation¹⁴⁰. Similarly, preliminary evidence suggests moderate efficacy of responsive neurostimulation, minimally invasive ablative strategies including laser interstitial therapy, and non-invasive neuromodulation in people with PTE^134,141.

Overall, we have a paucity of evidence to guide both medical and surgical care for PTE, with many patients not being referred to specialized care centres.

PTE biomarkers

Although most TBI centres do not screen for PTE, biomarkers hold promise to identify high-risk groups, with direct translational benefit for testing novel therapeutic targets. PTE might be unique among the clinical epilepsies in that it is attributable to a readily identifiable discrete event — that is, the TBI — which initiates epileptogenesis and confers an increased epilepsy risk. This characteristic facilitates the evaluation and identification of biomarkers at specific time points, ranging from hyperacute (minutes) to chronic (years). Two main categories of PTE biomarkers are commonly studied: clinical biomarkers comprising electrophysiological and radiological data, and molecular biomarkers, defined as protein and nucleic acid substrates.

Clinical biomarkers

One of the most intuitive biomarker categories for PTE is EEG. Acute EEG monitoring has become the standard of care for patients with TBI to identify EPTSs and status epilepticus. The presence of epileptiform abnormalities in the acute phase has been linked to increased risk of future development of PTE^142,143. Automated detection of these abnormalities further enables broad-scale utilization of these biomarkers¹⁴⁴. Findings from human retrospective data sets have been replicated in rodent models, providing additional insights into epileptogenesis^145–147. In two studies published in 2023, EEG background characteristics, as assessed by quantitative EEG, were also shown to aid PTE prediction in humans^142,148.

Structural brain imaging is a crucial component of early TBI evaluation, affording the evaluation of radiological characteristics as biomarkers of PTE. Multiple putative imaging biomarkers, including hippocampal volume, BBB disruption, diffusion tensor imaging and inflammatory and metabolic changes, have been reviewed previously^64,65. The Epilepsy Bioinformatics Study for Anti-Epileptogenic Therapy (EpiBioS4Rx) has elucidated several novel imaging biomarkers, including focal cortical and subcortical volume loss¹¹⁸, lesion location and differential volumetrics⁷⁴ and differential functional connectivity patterns¹¹⁶.

Molecular biomarkers

The unique nature of PTE as a model of epileptogenesis has engendered substantial interest in acute molecular biomarkers, given their potential translatability to other epilepsies. Neuronal, glial and axonal injury, along with inflammatory markers, have all been studied in TBI and previously reviewed¹⁴⁹. A systematic review identified blood neuregulin 1 and both blood and cerebrospinal fluid IL-1β as potential PTE biomarkers¹⁵⁰. Small-scale genetic association studies have also provided preliminary evidence for a genetic signature of PTE¹⁵¹.

MicroRNAs (miRNAs) — short non-coding RNA sequences that modulate protein synthesis and gene expression at the post-transcriptional level¹⁵² — have been identified as tractable targets for biomarker discovery in TBI and PTE. The broad effects of miRNAs on gene and protein expression, coupled with isoform-specific brain selectivity¹⁵² and peripheral bioavailability within serum exosomes, has led these RNAs to be implicated in a diverse array of neurological conditions¹⁵³. miRNAs have been evaluated as diagnostic biomarkers in a range of epilepsies^154–157, although current data regarding their predictive capacity in PTE are conflicting. The EPITARGET rodent study found that although neuronally enriched plasma miRNAs reflected the extent of cortical injury, they did not predict PTE development¹⁵⁸. By contrast, a systematic review evaluated the convergence of miRNA dysregulation in TBI and epilepsy and found 10 specific miRNAs (largely validated in serum) that were common to the two conditions, suggesting roles as PTE biomarkers¹⁵⁹.

Exosomes, which provide an important transport mechanism for miRNA, represent another novel class of PTE biomarkers. Exosomes are extracellular vesicles that are rich in bioactive molecules such as proteins and miRNAs, which modulate cellular environmental responses. They can be assayed in peripheral blood and might provide a window into CNS function. Exosome-based biomarkers, predominantly miRNAs, have been evaluated in both TBI¹⁶⁰ and epilepsy^{154,161–163}. Several groups have shown unique miRNA profiles in humans after TBI^164–169. Similarly, unique miRNA profiles have been linked to human epilepsy^155,170,171. Future work with peripheral exosomes in both preclinical animal models and people with PTE might produce new advances in PTE biomarker discovery.

An important caveat for studies evaluating candidate TBI biomarkers was raised by a preclinical study, which demonstrated that treatments such as prophylactic ASMs given in the acute post-TBI period, as well as EPTSs, can influence plasma protein biomarker levels and, potentially, their predictive value²³.

The gut microbiome

The gut microbiome has been explored for its role in the pathogenesis of brain disorders, including epilepsy. A preclinical study in the rat lateral fluid percussion injury model suggested that pre-existing gut microbiome could determine susceptibility to PTE¹⁷². A combination of microbiota species of the Lachnospiraceae family and short-chain fatty acids in the faeces predicted PTE with an area under the curve of 0.78 (ref. 172). Importantly, microbiome modification through microbiota transfer had positive effects on PTE risk¹⁷³.

Summary

PTE biomarkers have a vital role in the elucidation of the mechanisms underlying epileptogenesis and the development of anti-epileptogenic therapeutics. The role of these biomarkers in both diagnosis and measuring treatment response will be invaluable to these pursuits. Given the wide array of clinical and molecular biomarkers, we advocate the use of large-scale, collaborative, multimodal efforts, such as EpiBioS4Rx, for biomarker discovery and validation.

Epileptogenesis in PTE

Mechanisms

Epileptogenesis is the process by which a previously healthy brain tissue develops the propensity to generate spontaneous recurrent seizures. This process is well recognized in both clinical epilepsy and preclinical animal models^174,175. PTE provides a model disease condition not only for elucidating the underlying mechanisms but also for potentially halting epileptogenesis. TBI is a discrete inciting event, which is followed by a recognized latency period before the onset of clinical epilepsy (Fig. 2). In humans, the average latency period is several months¹¹ but can extend up to a decade¹⁷⁴. As in other epilepsies¹⁰⁸, epileptogenesis after TBI does not stop at clinical epilepsy onset; recurrent seizures continually remodel and reinforce seizure networks through strengthening of abnormal circuitry¹⁷⁶. This concept is consistent with clinical observations that epilepsy of an individual patient continues to evolve over their disease course, with changes in semiology, seizure frequency and pharmacoresistance.

Fig. 2 |. Mechanisms of epileptogenesis after traumatic brain injury.

Epileptogenesis is the chronic process whereby normally functioning brain tissue is altered to become prone to recurrent abnormal electrical activity that results in seizures. In the case of traumatic brain injury, the initial insult is followed by a latency period, during which a multitude of biological processes result in the formation of epileptogenic circuitry, ultimately leading to post-traumatic epilepsy (PTE). The epileptogenic process continues after seizure onset to reinforce and develop new abnormal circuits, causing the repetitive, spontaneous seizures that characterize PTE. AMPAR, AMPA receptor; NMDAR, NMDA receptor; NOS, nitric oxide synthase; ROS, reactive oxygen species.

A wide range of molecular mechanisms, including neuroinflammation, BBB dysfunction, cell signalling pathway perturbations, astrocyte dysregulation and epigenetic dysregulation, have been implicated in epileptogenesis in PTE. Ultimately, these molecular alterations lead to structural reorganization of the neural circuitry. These structural changes often have significant electrophysiological consequences, especially in PTE originating from the medial temporal lobe. In rodent models, hippocampal hyperexcitability early after TBI results from selective loss of inhibitory interneurons in the dentate gyrus and hilar regions^177–179. These interneurons help to mediate tonic inhibition through GABA activation of extrasynaptic receptors and are particularly susceptible to loss after TBI¹⁸⁰. Loss of these cells, as well as alterations in GABA receptor subunits^181,182, changes the inhibitory–excitatory balance, leading to reorganization of circuits¹⁸³. Reactive plasticity of excitatory dentate granule cells induces aberrant mossy fibre sprouting, connecting dentate granule cell axons with neighbouring neurons and creating recurrent excitatory circuits^184,185. In humans, a similar process might occur, with epileptogenesis further reinforcing epileptic circuits, thereby inducing neurodegeneration and gliosis to cause medial temporal PTE. About 30–60% of patients with PTE have seizures originating in the hippocampus^8,72.

Inflammation has long been considered as a key driver of epileptogenesis. Seizures and inflammation have a bidirectional relationship, with inflammation inducing seizures and vice versa. Work in animal models has shown that seizures activate nearby glial cells that subsequently release pro-inflammatory cytokines and mediators^5,183. As a source of secondary brain damage in the acute phase of TBI, seizures further exacerbate neuronal cell degeneration and initiation of the immune cascade. Within PTE, inflammation is a hallmark of secondary damage after TBI via glial cell activation and migration of inflammatory cells into the CNS⁵. Microglia are CNS-specialized macrophages, which, when chronically activated, produce cytotoxic molecules that contribute to oxidative stress, and phagocytose normal cells⁵. Astrocytes, which outnumber neurons at least tenfold in the human brain, maintain neuronal function through buffering of extracellular potassium and recycling of glutamate. Disruption of these functions leads to increased seizure susceptibility and alterations in excitatory circuitry. Astrocytes and microglia are abundant in post-surgical pathological specimens from patients with epilepsy, and they contribute to the recruitment of peripheral leukocytes and lymphocytes that are not typically found in the CNS^5,186. Increased activation of CNS-specific and peripheral inflammatory cells might partially explain the vulnerability of younger patients to PTE^11,29. Young animals and humans show an increased inflammatory responses to insults, such as altered microglial and cytokine expression patterns, that could increase PTE risk¹⁸³.

CNS and peripheral inflammatory cells release a plethora of inflammatory molecules that could contribute to PTE. Three signalling pathways, IL-1β, high-mobility group box 1–toll-like receptor 4 (HMGB1–TLR4) and C–C motif chemokine 2 (CCL2), might be drivers of epileptogenesis and are an important focus of research¹⁸³. IL-1β belongs to the IL-1 family of inflammatory cytokines, which mediate the innate immune response. IL-1β is produced by a range of lymphoid cells and initiates intracellular signalling with mitogen-activated protein kinase, nuclear factor ĸB and other factors to stimulate immune cell activation¹⁸⁷. IL-1β is highly expressed both immediately and several months after TBI and correlates with poor outcomes^188,189. Antibodies that neutralize this cytokine ameliorate oedema and cognitive deficits in mouse models of TBI¹⁹⁰. HMGB1 is a damage-associated molecule that is released by a wide range of CNS cells after tissue necrosis or damage¹⁹¹. HMGB1 was found at high levels in tissues resected during epilepsy surgery¹⁹² and probably contributes to epileptogenesis by increasing NMDA receptor function in TLR4-expressing hippocampal neurons¹⁹³. CCL2, a chemokine receptor that is involved in monocyte recruitment, is produced at high levels by astrocytes within hours of TBI¹⁹⁴. It is thought to modulate ion channel expression to induce a pro-excitatory state with decreased neuronal inhibition¹⁹⁵. Together, these pathways, along with other inflammatory molecules and cells, produce a pro-inflammatory environment that contributes to epileptogenesis. Although the work on these pathways was conducted in rodent models of epilepsy, surgical sampling of human epileptic tissue seems to confirm the active role of the immune system in epileptogenesis¹⁸⁶.

Another epileptogenic mechanism in PTE is BBB dysfunction, which occurs early after TBI, probably as a result of focal structural damage and changes in cerebrovascular permeability¹⁹⁶. In other epilepsies, BBB disruption is independently epileptogenic and levels of relevant biomarkers increase at seizure onset^197,198. In PTE, BBB disruption induces a pro-seizure state through altered potassium regulation caused by disruption of the potassium gradient, which is maintained by ATP pumps, and spatial buffering, which is maintained by astrocyte foot processes¹⁹⁹. The BBB can also be disrupted through alterations in the transforming growth factor β–albumin signalling pathway¹⁸³. As identified in animal models and verified in human studies, albumin induces a hypersynchronized excitatory response to electrical stimulation^200,201, with characteristic EEG alterations and paroxysmal slow cortical activity²⁰⁰. Albumin has been shown to bind to transforming growth factor β receptors on hippocampal neurons, inducing pro-excitatory cytokine production and increasing neuronal excitability¹⁸³.

Iron accumulation might also contribute to epileptogenesis in PTE. After TBI, intracranial bleeding and cell lysis increase extracellular iron deposition, thereby aggravating injury and inducing epileptogenesis²⁰². Excess iron generates free radicals, which exacerbate damage caused by reactive oxygen species released in the inflammatory cascade. In animal models, iron deposition induces epileptogenesis through upregulation of haem oxygenase 1, which initiates a cascade to release free iron. This process results in paroxysmal EEG discharges and seizure development²⁰³. Treatment with iron chelators might alleviate iron-associated damage and reduce PTE risk.

Therapeutic strategies

Alterations in specific signalling pathways after TBI might provide tractable therapeutic targets for PTE (Table 1). The mechanistic target of rapamycin (mTOR) signalling pathway has important roles in neuronal physiology, governing proliferation, growth, synaptic plasticity, morphology and cortical development^204,205. mTOR has been best studied in tuberous sclerosis, in which mTOR inhibition reduces seizure and tumour burden. In animal models of PTE, the mTOR inhibitor rapamycin also had a neuroprotective effect, attenuating post-traumatic neurogenesis, hippocampal mossy fibre sprouting and dentate granule cell excitability, but without a reduction in behavioural seizures²⁰⁶. WNT signalling may also have a role in epileptogenesis in PTE. This pathway affects neuronal signalling after TBI²⁰⁷ and was observed to be dysregulated in the hippocampus during epileptogenesis in preclinical animal models^208,209.

Table 1 |.

Therapeutic epileptogenic targets in post-traumatic epilepsy

Target	Mechanism	Potential therapies
Tau hyperphosphorylation	Hyperphosphorylation of tau protein to form p-tau is thought to impair neuronal cellular function and contributes to neurodegeneration202	Sodium selenate to reduce p-tau levels22,245,246
Oxidative damage caused by iron toxicity	Following brain haemorrhage, iron accumulation from haemolysis leads to generation of free radicals, mitochondrial fragmentation and oxidative damage	Iron chelator deferoxamine minocycline247
Neuroinflammation	Traumatic brain injury elicits a robust inflammatory response with microglial activation, disruption of the BBB, increased vascular permeability and release of inflammatory cytokines5,248	Anti-inflammatory agents including cyclooxygenase 2 inhibitors or targeted agents such as IL-1R1 antagonists249,250
BBB disruption	BBB disruption causes astrocyte dysregulation, as well as release of inflammatory macromolecules and leukocytes, ultimately resulting in progressive synaptic dysfunction	Dexamethasone; vascular endothelial growth factor inhibitors251
Epigenetic dysregulation	Post-transcriptional dysregulation of gene expression	Histone deacetylase inhibitors; DNA methyltransferase inhibitors; engineered site-specific chromatin remodellers252
Calcium channels	Disruption of high-voltage and low-voltage calcium channels alters intracellular calcium levels and induces neuronal death through apoptosis, burst firing from pyramidal cells and development of epileptogenic circuits	Drugs targeting calcium channels include ethosuximide, which is effective in animal models of absence epilepsy, and ziconotide, an N-type calcium channel blocker
mTOR	The mTOR signalling pathway governs neuronal proliferation, growth and morphology, synaptic plasticity and cortical development253,254	mTOR inhibitor rapamycin
WNT	WNT signalling is essential for the formation of neuronal circuits and for synaptic plasticity	WNT modulators209
Astrocyte dysregulation	Astrocytes assist in maintaining neuronal function by buffering extracellular potassium, recycling glutamate for neuronal signalling and regulating the excitatory-inhibitory transmission balance255	Modification of specific gene expression in astrocytes
Gut microbiome	The gut-brain axis refers to bidirectional communication between the enteric nervous system and the CNS172,173	Faecal transplantation

BBB, blood–brain barrier; mTOR, mechanistic target of rapamycin.

Modification of ion channels, in particular calcium channels, is another promising therapeutic strategy²⁰². Changes in intracellular calcium levels and calcium homeostasis have been linked to epilepsy in animal and human models²¹⁰. Several promising AEMs, some of which are currently used in clinical trials, target low-voltage T-type calcium channels that are widely distributed throughout the brain²¹¹.

Epigenetic dysregulation at various levels, including chromatin remodelling, histone post-translational modifications, non-coding RNAs and DNA methylation, has also been evaluated as a mechanism of epileptogenesis²¹², although it has not been studied extensively in PTE. Desbski et al. investigated DNA methylation patterns in three rodent models of epileptogenesis: a lateral fluid-percussion model, focal amygdala stimulation and systemic pilocarpine injection²¹³. They found both common epileptogenic and aetiology-specific DNA methylation signatures. Epigenetic mechanisms are particularly attractive therapeutic targets given the existence of established drugs such as histone deacetylase inhibitors and DNA methyltransferase inhibitors^214–216. Furthermore, technological advances have created the potential for more targeted epigenetic therapeutics, for example, engineered site-specific chromatin remodellers or CRISPR–Cas9 technology for targeted chromatin modifications^217,218.

Given the wide range of targetable mechanisms of epileptogenesis, PTE might serve as a model disease system for AEM development. PTE has a latency period from the initial injury, with epileptogenesis continuing after the first seizure, thereby providing an extended window for treatment. AEMs have the potential to prevent, stop or reverse the development of epilepsy. The ideal AEM would be administered over a molecularly relevant and clinically recognizable treatment window, with durable effects beyond the treatment period and without long-lasting adverse effects²¹⁹. By contrast, conventional ASM therapy is aimed at reducing seizure burden but not necessarily halting epileptogenesis¹³³. Traditional ASMs have been trialled as anti-epileptogenic agents in humans; however, attempts to repurpose these medications for PTE prevention have largely failed^85,95.

No broad-spectrum AEMs are currently clinically available but they are the subject of intensive research in the preclinical arena. Several large multi-institutional consortiums, including the Team Approach to the Prevention and Treatment of PTE (TAPTE), EpiBioS4Rx and EPiXchange, have created strong networks to share ideas and data to develop bench-to-bedside anti-epileptogenic treatments. These groups and others have developed various preclinical models of epileptogenesis, including blunt or focal cortical injury, in a range of animals, from mice to pigs²²⁰. These research efforts have led to exciting preclinical proposals for AEMs and other interventions to modify epileptogenesis.

Although no candidate drug for PTE has so far been validated for human use, this condition has far-reaching potential as a model of epileptogenesis. Successful AEM development in the context of human PTE might not only improve global outcomes in TBI but also inform novel therapies that are broadly applicable to all epilepsies.

Conclusions and future priorities

In this Review, we have presented the clinical complexities of PTE and highlighted the role of PTE research in elucidating the mechanisms of epileptogenesis. PTE is common after TBI and is associated with markers of injury severity, including craniotomy and comatose state. Despite its high prevalence and association with adverse neurological outcomes, treatment strategies for PTE are variable and poorly defined. Promising research has identified clinical, EEG, imaging and fluid biomarkers that might aid stratification of PTE risk.

Building on findings from multi-institutional working groups, we propose that PTE is an appropriate model of human epileptogenesis, affording the discovery of both novel biomarkers and treatment strategies, which might be applicable to epilepsy in general. We have highlighted several potential epileptogenic mechanisms in PTE, including neuroinflammation, BBB dysfunction, cell signalling pathway perturbations, astrocyte dysregulation and epigenetic dysregulation. Future prospective, multicentre, collaborative efforts should uncover novel epileptogenic biomarkers, mechanisms and therapies. Successful treatment strategies for PTE will extend the treatment options from symptomatic therapies to disease-modifying anti-epileptogenic therapies, reducing both the development and progression of PTE.

Successful deployment of anti-epileptogenic therapies will rely heavily on appropriate clinical trial design. Classic RCT designs minimize bias and confounding but are expensive and have limited real-world applicability. When considering RCTs for PTE therapeutics, biomarker discovery — both clinical and molecular — is paramount to enable the recruitment of specific high-risk TBI populations, thereby considerably reducing sample sizes. As participants enrolled in this type of study require long-term follow-up for the primary outcome of PTE development, sample size optimization will be invaluable to trial success. Given the established drawbacks of RCTs, other innovative trial designs might need to be considered.

After a long list of failed clinical trial interventions to treat acute TBI and the lack of FDA-approved therapies for TBI, several funders have adapted their paradigms for clinical research funding within the area of acute brain injury and its sequelae. Some have suggested increased use of pragmatic clinical trials^221,222 (PCTs) — a strategy that is intended to reflect real-world implementation of a new intervention. In contrast to the explanatory premise of traditional RCTs, PCTs aim to model future therapeutic effectiveness using high-quality real-world evidence^223,224. Although attractive in concept, however, PCTs might not address all the needs of the field. In one example of a pragmatic trial design, 387 patients were recruited from real-world clinical practice settings across 47 centres in 18 countries to test the effectiveness of therapeutic hypothermia to reduce intracranial pressure after TBI²²⁵. Notably, the nature of the treatment did not allow blinding, and the trial failed to demonstrate an effective functional outcome.

The need for community and stakeholder engagement is a key issue for future PCTs^25,226,227. To ensure treatment equity, engaging individuals from socially disadvantaged, underserved and minoritized backgrounds will be vital to the success of prospective studies and translation to trials and future practice^228–230. Health inequities have been observed at multiple levels among critically ill individuals, including access to care, acute care and post-hospital care outcomes^231–233. Effective trial design will require community input representing groups that are affected both by PTE itself and by future practice changes that might emerge from the study^234–236. These approaches have been more extensively explored and implemented in the field of oncology and social medicine, which might help to inform the TBI, epilepsy and critical care research fields^237,238.

To successfully develop PTE clinical trials, investigators will need to better understand and define the desired outcomes. Participatory action research (PAR) and community engagement approaches have been used to broaden our understanding of TBI rehabilitation outcomes and long-term comorbidities, to engage with rural communities on their perspectives of neurological outcomes and to evaluate potential participation in long-term outcome studies among individuals living with post-TBI sequelae^239–241. In addition, some centres used a PAR approach in pharmacological RCTs for post-acute brain injury management^242–244. True community engagement and PAR should involve patient and community stakeholders from the time of study conceptualization and design to ensure adequate consideration of and input into aspects such as budgeting for further community engagement, institutional review board approval for engagement and perspectives from people and families involved in the study and to devise a plan for public education and knowledge transfer to the affected communities.

We are optimistic for the future of PTE research and treatment. A combination of robust clinical and basic science research initiatives will continue to improve current clinical practice and inform future treatments, and PTE research also holds great promise for anti-epileptogenic therapy discovery and utilization.

Key points.

• Post-traumatic epilepsy (PTE) is highly prevalent after traumatic brain injury, impairing neurological recovery and leading to worse functional outcomes.

• Current epilepsy therapeutics symptomatically treat seizures but do not modify epileptogenesis, the process by which brain tissue becomes prone to seizures.

• The unique nature of PTE, occurring after a well-defined epileptogenic insult, makes it a promising model system for understanding epileptogenesis.

• Future research in individuals with PTE populations might not only reveal novel mechanisms of epileptogenesis but also enable anti-epileptogenic therapies to be tested.

Glossary

Anti-epileptogenic medications

(AEMs). Therapies that ameliorate epileptogenesis, with lasting effects beyond the period of drug exposure.

Anti-seizure medications

(ASMs). Medications that treat seizures but do not modify the process of epileptogenesis or alter the disease course of epilepsy.

Cerebrospinal fluid shunting

A common neurosurgical procedure to drain cerebrospinal fluid, usually from the ventricular system, thereby decreasing intracranial pressure.

Encephalomalacia

A radiological finding denoting an area of brain tissue that has undergone liquefactive necrosis.

Engel Epilepsy Surgery Outcome Scale

A classification scheme of seizure outcomes after epilepsy surgery using four classes: 1, free of disabling seizures; 2, rare disabling seizures; 3, worthwhile improvement; and 4, no worthwhile improvement.

Extended GOS

(GOSE). An extension of the Glasgow Outcome Scale that subdivides the categories of severe disability, moderate disability and good recovery into lower and upper categories.

Glasgow Coma Scale

(GCS). A broadly utilized clinical scale describing the level of consciousness after traumatic injury.

Glasgow Outcome Scale

(GOS). A global scale for functional outcome after brain injury that rates patient status using five categories: dead, vegetative state, severe disability, moderate disability and good recovery.

Mild TBI

Traumatic brain injury with post-impact (may not need resuscitation) Glasgow Coma Scale score 13–15.

Moderate TBI

Traumatic brain injury with post-resuscitation Glasgow Coma Scale score 9–12.

Pathological posturing responses

Stereotypical movements of the trunk and extremities in response to stimuli, typically indicative of significant CNS injury.

Severe TBI

Traumatic brain injury with post-resuscitation Glasgow Coma Scale score ≤ 8.