Diaz-Arrastia R, Agostini MA, Frol AB, Mickey B, Fleckenstein J, Bigio E, Van Ness PC. Neurophysiologic and Neuroradiologic Features of Intractable Epilepsy After Traumatic Brain Injury in Adults. Arch Neurol. 2000;57(11):1611–1616. doi:10.1001/archneur.57.11.1611
There is controversy regarding the precise mechanism by which epilepsy results after traumatic brain injury (TBI). Previous reports have suggested that mesial temporal lobe epilepsy may result from TBI only in young children, while neocortical epilepsy arises from TBI in later life. These conclusions were based on surgical series and may be biased because of patient selection.
To determine the frequency of mesial temporal lobe as opposed to neocortical epilepsy in patients with intractable epilepsy resulting from TBI after the age of 10 years.
Patients and Methods
We identified 23 patients with intractable epilepsy who had TBI after the age of 10 years, preceding the onset of epilepsy. Patients were studied by simultaneous videotape and scalp electroencephalographic recording of typical seizures; magnetic resonance imaging; neuropsychologic studies; and, when appropriate, intracarotid amobarbital testing. Two patients underwent anterior temporal lobectomies.
Of the 23 patients, 8 (35%) had mesial temporal lobe epilepsy, based on the finding of hippocampal sclerosis on a magnetic resonance imaging scan, consistent interictal and ictal electroencephalographic recordings, evidence of temporal lobe dysfunction on neuropsychologic testing, and characteristic seizure semiology. Two of these patients underwent anterior temporal lobectomies with clinical benefit, and hippocampal sclerosis was confirmed pathologically. In 2 cases, patients were not treated surgically because of bilateral temporal lobe dysfunction noted on intracarotid amobarbital testing. Eleven patients had neocortical epilepsy; 1 had primary generalized epilepsy; and, in 3, the site of seizure onset was not localized.
Mesial temporal lobe epilepsy can result from TBI in adolescents and adults as well as in children, and can often be bilateral and associated with multifocal injury. This information may be useful in developing prophylactic therapy for posttraumatic epilepsy.
TRAUMATIC brain injury (TBI) is a common cause of epilepsy, accounting for approximately 4% of focal epilepsy in the general population, and is the leading cause of epilepsy with onset in young adults (those aged 15-24 years).1 Epilepsy resulting from brain trauma is often difficult to control with medical therapy, and is the cause of epilepsy in approximately 5% of patients referred to specialized epilepsy centers.2
Traumatic brain injury results in potentially epileptogenic brain damage through several mechanisms, which often coexist within a single patient. Penetrating brain injury produces a cicatrix in the cortex and is associated with a risk of posttraumatic epilepsy (PTE) of approximately 50%.3 Nonpenetrating head injuries may produce focal contusions and intracranial hemorrhages, and are associated with a risk of PTE of up to 30%.4 In this setting, the mechanism of epileptogenesis may be partly related to the toxic effects of hemoglobin breakdown products on neuronal function.5 Finally, closed head injury often produces diffuse concussive injury, with shearing of axons, diffuse edema and ischemia, and secondary cellular damage through the release of excitatory amino acids, cytokines, bioactive lipids, or other toxic mediators.6 The incidence of PTE after diffuse head injury is less established, but is likely to be in the order of 10% in patients who had loss of consciousness for longer than 24 hours.4 Epileptogenesis may arise from diffuse injury as a result of selective damage to vulnerable brain regions, such as the hippocampus.7
Several workers8- 10 have studied brain trauma as the cause of intractable temporal lobe epilepsy retrospectively in surgical series of patients who underwent anterior temporal lobectomy (ATL). These studies concluded that TBI could result in mesial temporal lobe epilepsy (MTLE) when injury occurred early in life (age <5 years).8,9 These results have been interpreted as evidence of a special vulnerability of the developing brain, which makes TBI in children more likely to result in hippocampal sclerosis and intractable MTLE than injury later in life.8,11 These studies had selection bias (only patients judged to be good candidates for surgical therapy after extensive evaluation were included), and are unlikely to reflect the full spectrum of partial epilepsy after head injury. Thus, we decided to study the clinical, electrophysiologic, and neuroradiologic features of patients referred to our center for videotape and electroencephalographic (EEG) monitoring who had moderate to severe head injury after the age of 10 as a risk factor for intractable epilepsy.
The records of 474 consecutive admissions to the Epilepsy Monitoring Unit at Parkland Memorial Hospital, Dallas, Tex, from May 1998 to December 1999, were reviewed. Patients and family members were questioned extensively on multiple occasions by attending staff (R.D.-A., M.A.A., and P.C.V.N.), house staff, and nursing staff as to possible risk factors for epilepsy, and the data were recorded in the medical records. All credible risk factors were recorded, including history of head injury, febrile seizures, meningitis or encephalitis, significant perinatal insult, or a family history of epilepsy. Patients were included if they had sustained moderate to severe head injury12 that preceded the onset of epilepsy and was of sufficient magnitude to result in prolonged (≥30 minutes) loss of consciousness or amnesia, hospitalization, or neuroradiologic evidence of TBI. Patients with other risk factors for epilepsy, such as febrile convulsions, meningitis, or encephalitis, were excluded. Thirty-three patients (7.0%) who met the inclusion criteria were identified. Of these, 23 had diagnostic videotape-EEG evaluations, and were included in the study. Nine had no seizures during their admission, and 1 had only nonepileptic seizures.
All patients underwent prolonged surface ictal and interictal audiovisual and EEG recording. Data were collected (BMSI 5000 system; Nicolet Miomedical, Milwaukee, Wis), at a 400-Hz sampling rate, using the international 10-20 and modified combinatorial nomenclature system of electrode placement. Ictal events were identified by the patient, by family members, and by computerized spike detection algorithms. Computerized spike and seizure detection was done nightly. Interictal and ictal EEGs were reviewed by board-certified electrophysiologists (M.A.A. and P.C.V.N.). Detailed descriptions of seizure semiology were recorded.
All patients underwent magnetic resonance imaging (MRI) of the brain, using a technique sensitive for detecting sclerosis of the mesial temporal structures. T1-weighted sagittal and gradient echo axial images were obtained, in addition to fluid-attenuated inversion recovery and T1- and T2-weighted coronal images obtained at a 3-mm-slice thickness through the region of the hippocampus. Mesial temporal sclerosis (MTS) was defined by the finding of hippocampal atrophy, T2-weighted hyperintensity, or a bright fluid-attenuated inversion recovery signal, all assessed by visual inspection. All scans were reviewed by board-certified neuroradiologists (including J.F.) who were blinded to patient risk factors. As others13 have reported, in our experience visual inspection by qualified observers is comparable to quantitative assessment of hippocampal volume or T2 relaxation times.
Of the 23 patients, 9 were tested with a comprehensive neuropsychologic battery consisting of the following measures: intelligence—Wechsler Adult Intelligence Scale, revised or third version; academic—Wide Range Achievement Test 3; executive functions—Wisconsin Card Sorting Test, Ruff Figural Fluency Test, and Trail-Making Test; attention—Digit Span (Wechsler Adult Intelligence Scale), Digit Vigilance Test, and Trail-Making Test (part A); language—Boston Naming Test, Oral Fluency Test (FAS [standard word fluency test] or animals), and Vocabulary (Wechsler Adult Intelligence Scale); visuoconstruction—Block Design (Wechsler Adult Intelligence Scale), Clock or Cross Drawing, and Ray-Osterrieth Complex Figure copy; memory—California Verbal Learning Test, Ray-Osterrieth Complex Figure recall, Warrington Recognition Memory Test, and selected tests from the Wechsler Memory Scale (revised or third version); and motor—finger tapping or grip strength.
Two patients (patients 1 and 2) underwent ATLs. Pathological examination results of en bloc resections were reviewed by one of us (E.B.), a board-certified neuropathologist who was blinded to patient risk factors and surgical outcome. Patients were evaluated for the possibility of classic hippocampal sclerosis according to the criteria developed by Margerison and Corsellis14 and Rushing et al,15 consisting of neuron loss in CA1 and CA4 and the presence of associated reactive gliosis, with variable loss of dentate granule neurons. Because of fragmentation and incomplete removal of the amygdala, we chose to limit the evaluation of neuronal loss and gliosis to CA1 and CA4.
A total of 474 patients were evaluated in our epilepsy monitoring unit during the study period, of which 247 (52.1%) had videotape-EEG evaluations diagnostic of epilepsy. The remaining 227 (47.9%) had nondiagnostic evaluations or were diagnosed as having exclusively nonepileptic seizures. Of the total group, 33 (7.0%) had moderate to severe TBI after the age of 10 years, which had preceded the onset of unprovoked seizures. Of the cohort of 33 patients, 23 (70%) had diagnostic videotape-EEG evaluations. The other 10 had no seizures recorded (9 patients) or had only nonepileptic seizures (1 patient). The clinical characteristics of the 23 patients who underwent diagnostic evaluations are outlined in Table 1. There were 13 men and 10 women. The mean ± SD age of the patients at the time of injury was 22.9 ± 9.4 years (median, 21 years), the mean ± SD age at onset of refractory epilepsy was 24.6 ± 9.3 years (median, 23 years), and the mean ± SD duration from the time of injury to the onset of refractory epilepsy was 1.6 ± 2.3 years (median, <12 months). The short time from injury to onset of habitual seizures supports a causative relation between TBI and epilepsy in these patients. The mean ± SD duration from the onset of epilepsy to referral for videotape-EEG monitoring was 10.9 ± 9.5 years (median, 8 years). All patients had medically refractory epilepsy, defined as having frequent uncontrolled seizures despite adequate trials of at least 2 antiepileptic medications. The results of interictal and ictal EEG recordings, MRI scans, neuropsychologic evaluation, intracarotid amobarbital testing, and surgical pathological findings are summarized in Table 2.
Of the 23 patients who underwent diagnostic evaluations, 8 (35%) had MTLE. The diagnosis can be considered definitive only in 2 patients who underwent ATLs and had a significant reduction in the frequency of their seizures. Patient 1 is seizure free 6 months after surgery, and the other patient (patient 2) had a greater than 90% reduction in seizure frequency, although he still has monthly seizures. Classic hippocampal sclerosis, consisting of neuronal loss and gliosis in CA1 and CA4, was found in both resected hippocampi. In the other 6 patients, the diagnosis of MTLE, while not definitive, is highly probable based on the following: (1) all had radiologic evidence of atrophy, T2 signal shortening on high-resolution MRI, or both; (2) all had exclusively temporal lobe interictal epileptiform discharges; (3) all had seizure semiology consistent with mesial temporal lobe onset of their seizures; (4) all had an ictal EEG showing seizure onset in the temporal lobes16,17; and (5) neuropsychologic evaluations in tested patients were indicative of temporal lobe dysfunction. The combination of these criteria is widely accepted as a reliable approach that predicts success after temporal lobectomy, and is highly correlated with finding MTS pathologically.16 Of the 6 patients who have not been operated on, 2 were turned down for surgery because of bilateral temporal lobe memory deficits identified on intracarotid amobarbital testing.
To establish a basis for comparison, we reviewed the causative factors of all 247 patients who had diagnostic videotape-EEG evaluations during the study period. Table 3 shows the frequency of an MTLE diagnosis as a function of presumptive cause. Mesial temporal lobe epilepsy was diagnosed in approximately one fourth of the patients. We were unable to identify a clear causative factor in slightly more than half of our patients, and the frequency of MTLE in these cryptogenic cases was comparable to that of the group as a whole. As others11,18,19 have found, a history of febrile seizures is preferentially associated with the diagnosis of MTLE. We also find that focal neurologic insults, such as result from tumors, vascular malformations, infarcts, or brain hemorrhages, rarely lead to MTLE. The likelihood of diagnosing MTLE in patients with a history of TBI (at any age), 31.1%, is comparable to that found in the cryptogenic cases (25.1%) and to that found in patients with other diffuse causes, such as meningitis or encephalitis (29.4%).
In the United States, approximately 350,000 patients are hospitalized each year with TBI, in addition to 1.5 million cases of milder TBI that do not require hospitalization.20 A major source of morbidity after TBI is PTE, which can complicate 25% to 30% of cases of severe head injury and 5% to 10% of cases of mild to moderate injury.21 Posttraumatic epilepsy is the most common cause of new-onset epilepsy in young people,1 is often refractory to medical therapy, and is the cause of epilepsy in approximately 5% of patients referred to specialized epilepsy centers.2
Because of the social burden of PTE, attempts have been made to develop prophylactic antiepileptogenic therapy, but rigorous trials22,23 have so far been unsuccessful. Recently developed anticonvulsant drugs have neuroprotective properties24- 26 but have not yet been tested for antiepileptogenic properties in either animals or humans. A major impediment to developing effective prophylactic therapy is the absence of a well-accepted animal model that reliably reflects the synaptic and cellular events that result in PTE. Kindling has been proposed as a potential animal model to screen for the ability of drugs to prevent PTE in humans,27 while others5,28 have emphasized animal models involving focal injury in the cortex, in which surgical lesions, blood, or iron injections result in neocortical epilepsy. It is likely that the molecular and cellular events that result in epileptogenesis differ in these models, and that each may partially mimic posttraumatic epileptogenesis in humans. Successful development of prophylactic antiepileptogenic therapy will depend on detailed understanding of the mechanisms responsible for human PTE.
In humans, direct injury to the hippocampus from TBI is uncommon. Courville29 examined the brains of 108 patients who had fatal TBI and found contusions in the hippocampus in only 11 (10.2%). Other pathologic series of fatal TBI30,31 have found neuronal loss primarily in the CA1 subfield of the hippocampus, which was frequently bilateral. Presumably hippocampal sclerosis and MTS result from diffuse, secondary effects of TBI.
More recently, several groups8,9,19 have studied patients who underwent ATL as therapy for refractory epilepsy. Mathern et al8 studied 259 patients who underwent ATL at the University of California, Los Angeles, from 1961 to 1992, 26 (10%) of whom had TBI as the major risk factor. They found that 50% of these patients had hippocampal sclerosis, a frequency not significantly different from patients in their series who had other risk factors or cryptogenic temporal lobe epilepsy. These workers emphasized that the mean ± SD age at the time of injury in these patients was 6.3 ± 1.6 years (median, 2.5 years), while the mean ± SD age of onset of habitual temporal lobe epilepsy was 18.1 ± 1.8 years. Marks et al9 described 25 patients with PTE who were examined at Yale University, New Haven, Conn, from 1982 to 1992, 21 of whom were treated surgically. Seventeen of those patients were judged to have MTLE, and 14 were treated with ATL. Of these, 6 (35%) had hippocampal sclerosis confirmed pathologically and had excellent outcomes postoperatively. Of the remaining 8 patients who were operated on, none had a favorable outcome. These researchers emphasized that all patients with hippocampal sclerosis were younger than 5 years (mean age, 3.4 years) at the time of initial injury. The Yale group also found 8 patients with neocortical epilepsy, 5 of whom were treated with focal cortical resections. Only 3 patients had good outcomes after surgery, and all 3 had a circumscribed focal lesion noted on an MRI scan. Patients with neocortical epilepsy were significantly older (average age, 18.25 years) at the time of the initial injury. These researchers concluded that MTS results from injury early in development, and that injury later in life is less likely to produce hippocampal sclerosis. A third surgical series19 described 102 patients who underwent ATL at the University of Michigan, Ann Arbor, from 1990 to 1996. Twenty-nine (28.4%) had head trauma as a cause, of which 20 (69%) had hippocampal sclerosis identified pathologically. In contrast to the previous series,8,9 this study did not find a correlation between age at the time of injury and the presence of MTS or seizure control after surgery.
These earlier reports focused on highly selected patients who underwent extensive scalp and intracranial ictal EEG recordings, neuroimaging, and neuropsychologic evaluations, and were believed to be good candidates for resective surgery. Patients with PTE who had poorly localized, multifocal, and nonlesional extratemporal epilepsy were likely excluded from these series. Thus, we decided to study patients referred to our center who had a history of TBI preceding the onset of intractable epilepsy. We elected to focus on patients who had TBI after the age of 10 years because of suggestions that the susceptibility of the hippocampus and mesial temporal structures to injury was restricted to the first 5 years of life.9
Our finding of MTS, as identified by neuroimaging and neurophysiologic techniques (6 patients) and confirmed pathologically (2 patients), suggests that TBI can result in hippocampal injury in adults, supporting the conclusions of Schuh et al.19 It is likely that differences in referral patterns and patient selection account for the failure to find MTS in patients with injury at older age groups in the studies from the University of California, Los Angeles,8 and Yale University.9 Since TBI is most common in young adults, this finding has implications for the development of therapeutic interventions to prevent PTE. Because of the retrospective nature of our study, it is impossible to exclude the possibility that these patients had preexisting, clinically silent hippocampal sclerosis, and that epilepsy was expressed only after injury. We believe that this explanation is unlikely since hippocampal sclerosis is rare in nonelderly patients who do not have temporal lobe epilepsy.32,33 It is possible, however, that patients who develop MTS after head injury have a genetic predisposition to hippocampal injury, the nature of which is so far unknown. Epidemiologic support for this possibility is the fact that patients with PTE frequently have a family history of epilepsy.34,35
Two patients underwent ATL and had a beneficial outcome. One patient with radiologically evident MTS had nonlocalizing ictal EEG recordings, and is potentially a candidate for surgical resection, but has elected to defer intracranial EEG evaluation. Another 3 patients in our series had MTLE and radiologically apparent MTS, but were not candidates for ATL. In 2 patients, this was because of severe bilateral temporal lobe memory dysfunction, confirmed by intracarotid amobarbital testing. These findings are consistent with the conclusions of Kotapka et al,31 who determined pathologically that hippocampal neuron loss after TBI was often bilateral. Another patient had radiologic evidence of hippocampal sclerosis and an extensive extratemporal lesion. Ictal EEG studies in this patient were not localizing as to the precise site of seizure onset. Since focal and diffuse head injury may coexist in a single patient, the finding of dual pathological features is not surprising. Our findings support the hypothesis that MTS can result from TBI in adolescents and adults, but also is frequently associated with bilateral temporal lobe dysfunction, coexisting neocortical lesions, or both. Although our numbers are small, our experience is consistent with the findings of Schuh et al,19 who found that a history of head trauma decreased the likelihood of a good outcome after ATL, even in patients with pathological evidence of hippocampal sclerosis in the resected specimen.
Three patients had focal neocortical encephalomalacia on MRI scans, which likely resulted from old contusions or hematomas. These patients were not believed to be candidates for surgical treatment because of the large and multilobar area of injury (2 patients) and a coexisting frontal neocortical lesion and hippocampal sclerosis (1 patient). Our impression is in agreement with Marks et al,9 who believed that most patients with neocortical PTE were not good candidates for surgical therapy. These disappointing results emphasize the need for effective prophylactic antiepileptogenic therapy. One patient (patient 23) met all criteria for inclusion in the study, but had idiopathic generalized epilepsy, most likely unrelated to her severe head injury.
In summary, our series found that MTLE and neocortical epilepsy can result from TBI in adolescents and adults. While many patients with PTE can be successfully treated with resective surgery, TBI often results in bilateral and multifocal injury, and preoperative evaluation of these patients should be mindful of this fact. Our findings add potentially important information to understanding the process of epileptogenesis after TBI in humans, a common clinical problem that awaits the development of effective therapy.
Accepted for publication April 28, 2000.
This study was supported by grants KO8 NS01763, RO1 AG12297, and RO3 AG16450 from the National Institutes of Health, Bethesda, Md (Dr Diaz-Arrastia).
Reprints: Ramon Diaz-Arrastia, MD, PhD, Department of Neurology, The University of Texas Southwestern Medical Center at Dallas, 5323 Harry Hines Blvd, Dallas, TX 75235-9036 (e-mail: RdiazA@mednet.swmed.edu).