Article Text

Download PDFPDF

Blood–brain barrier disruption in post-traumatic epilepsy
  1. O Tomkins3,
  2. I Shelef2,
  3. I Kaizerman3,
  4. A Eliushin2,
  5. Z Afawi4,
  6. A Misk5,
  7. M Gidon2,
  8. A Cohen2,
  9. D Zumsteg6,
  10. A Friedman1
  1. 1
    Departments of Physiology and Neurosurgery, Soroka Medical Center and Zlotowski Center for Neuroscience, Ben-Gurion University of the Negev, Beer-Sheva, Israel
  2. 2
    Department of Neuroradiology, Soroka Medical Center, Beer-Sheva, Israel
  3. 3
    Department of Physiology Zlotowski Center for Neuroscience, Ben-Gurion University of the Negev, Beer-Sheva, Israel
  4. 4
    Department of Neurology, Tel-Aviv Sourasky Medical Center, Tel Aviv, Israel
  5. 5
    Department of Neurology, Shaare Zedek Medical Center, Jerusalem, Israel
  6. 6
    Department of Neurology, University Hospital Zurich, Zurich, Switzerland
  1. Alon Friedman, Departments of Physiology and Neurosurgery, Zlotowski Center for Neuroscience, Ben-Gurion University, 84105 Beer-Sheva, Israel; alonf{at}


Background: Traumatic brain injury (TBI) is an important cause of focal epilepsy. Animal experiments indicate that disruption of the blood–brain barrier (BBB) plays a critical role in the pathogenesis of post-traumatic epilepsy (PTE).

Objective: To investigate the frequency, extent and functional correlates of increased BBB permeability in patient with PTE.

Methods: 32 head trauma patients were included in the study, with 17 suffering from PTE. Patients underwent brain MRI (bMRI) and were evaluated for BBB disruption, using a novel semi-quantitative technique. Cortical dysfunction was measured using electroencephalography (EEG), and localised using standardised low-resolution brain electromagnetic tomography (sLORETA).

Results: Spectral EEG analyses revealed significant slowing in patients with TBI, with no significant differences between patients with epilepsy and those without. Although bMRI revealed that patients with PTE were more likely to present with intracortical lesions (p = 0.02), no differences in the size of the lesion were found between the groups (p = 0.19). Increased BBB permeability was found in 76.9% of patients with PTE compared with 33.3% of patients without epilepsy (p = 0.047), and could be observed years following the trauma. Cerebral cortex volume with BBB disruption was larger in patients with PTE (p = 0.001). In 70% of patients, slow (delta band) activity was co-localised, by sLORETA, with regions showing BBB disruption.

Conclusions: Lasting BBB pathology is common in patients with mild TBI, with increased frequency and extent being observed in patients with PTE. A correlation between disrupted BBB and abnormal neuronal activity is suggested.

View Full Text

Statistics from

Traumatic brain injury (TBI) is considered to be a major risk factor for post-traumatic epilepsy (PTE).1 The mechanisms underlying the development of PTE and its prevention remain unknown.2 Recent animal experiments show that opening the blood–brain barrier (BBB) leads to network changes, long-lasting epileptiform activity and eventual neurodegeneration.3 4 Although previous clinical studies show that altered permeability is observed in neurological patients, including those with TBI,57 no data exists regarding the frequency, extent and significance of BBB disruption in PTE. The aim of this study was to examine BBB permeability following TBI and to correlate it with electroencephalography (EEG) abnormalities and the presence of PTE.


Patient selection

The study protocol was approved by the Soroka Medical Center Helsinki Committee. Seventeen patients diagnosed with PTE were included. All except one reported repeated partial seizures (with secondary generalisation in 10 patients). The control group included 15 patients with non-epileptic TBI who reported headaches (86.7%), cognitive impairment (6.7%) or motor dysphasia (6.7%) (table 1). Patient groups were similar in age (27.9±4.2 and 25.3±2.8 years, respectively, p = 0.97) and gender (7 women and 10 men, and 4 women and 11 men, respectively, p = 0.47). All patients were healthy prior to TBI without history of neurological or psychiatric disorders. Most patients (90.6%) suffered mild TBI according to a Glasgow Coma Score (GSC) of >13 following admission. Of all the patients with TBI, 13 were unconscious for several minutes and 3 for several days, with no differences between the groups.

Table 1 Patient characteristics


EEG recordings were made by a clinical 128-channel digital EEG acquisition unit (CEEGRAPH IV, Bio-logic Systems Corp., Mundelein, Illinois, sampling rate of 256 Hz), as previously reported by our laboratory.7 The EEG was interpreted by a physician who was unaware of the study. The average power for the discrete frequency bands was normalised to each subject’s own total power (1.5–40 Hz). Recordings of 13 healthy adult volunteers (aged 34.8±2.8 years), with no history of brain injury or neurological disease, served as controls. For localisation, sLORETA was used.7 8

MRI and BBB integrity evaluation

MRI scans were performed using a 1.5 tesla machine (Intera, Philips Medical Systems, Best, The Netherlands). Scans were only performed after 24 hours without seizures. For the evaluation of BBB integrity, images were collected before and after peripheral administration of the contrast medium Magnetol (Gadolinium-DTPA (Gd-DTPA) 0.5 M, 0.1 mmol/kg) (Soreq Radiopharmaceuticals, Israel). BBB permeability was estimated as previously reported.5 6 In short, axial T1-weighted spin-echo images (582/15/1 [TR/TE/NEX], section thickness, 5 mm; intersection gap, 1 mm; matrix, 256 × 256) were obtained. Matching brain images from before and after Gd-DTPA administration were paired and analysed for statistically significant changes in signal intensity. For lesion volume measurements, T1 MRI images before contrast agent administration were used. The intracerebral lesion was identified by a physician and the number of pixels within the lesion was counted in all slices. Disruption volume was calculated for the same slices by counting all pixels with significant enhancement. Localisation of the cortical and BBB lesions according to Brodmann areas was performed by manual anatomical registration to the digitised Talairach brain atlas.


EEG recordings were performed on 22 patients, 10 days–11 years (median = 5.5 months) after the trauma. Apparent slowing or interictal epileptiform activity was observed in 78.6% of patients with PTE and in 12.5% of patients without epilepsy (p = 0.006, χ2 Pearson’s test) (table 1). Spectral analysis showed that the delta (1.5–4 Hz) power in both PTE (3.68±0.28%) and non-epileptic groups (3.76±0.29%) was significantly higher than that of healthy controls (2.83±0.15%, p<0.01, Mann–Whitney U test). Only the PTE group showed significantly elevated theta and reduced alpha (2.28±0.14 vs 1.93±0.09%, p = 0.04 and 1.99±0.14 vs 2.83±0.15%, p = 0.01, respectively, Mann–Whitney U test) compared with controls. No significant differences were found between the PTE and non-epileptic groups (data not shown).

The MRI scans of all patients were evaluated for intracerebral lesion and BBB disruption volumes. In 14 (56%) patients with TBI, brain regions were identified with significant enhancement, indicating increased BBB permeability (fig 1A). In 13 patients (86.7%), the disrupted BBB was in close proximity to cortical regions around old haemorrhagic contusions. Only one patient displayed BBB disruption with no concomitant lesion. BBB disruption was identified months or years (21.8±10.6, median  =  2 months) after the trauma (table 1). Parenchymal lesions were found in all cortical regions (table 1). The average lesion volume was 6.0±1.7 cm3 and the average volume of cortex with abnormal BBB was 5.9±1.6 cm3. Although patients with PTE were more likely to have a lesion on their MRI scans (80%) than patients without epilepsy (30.8%, p = 0.02, χ2 Pearson’s test, fig 1A), there was no significant difference in the size of the lesion (6.6±1.9% vs. 5.3±2.8 cm3, p = 0.19, Mann–Whitney U test, fig 1B). Patients with PTE were more likely to have BBB disruption (76.9%) than patients with non-epileptic TBI (33.3%, p = 0.047, χ2 Pearson’s test), and the volume of the BBB-disrupted cortex was significantly larger (9.8±2.6 vs 1.7±0.6 cm3, p = 0.001, Mann–Whitney U test, fig 1B).

Figure 1 (A) Statistically significant enhancement of T1 MRI scan in the region surrounding the cortical lesion in patient no 21, 10 days following the trauma. (B) Among patients with post-traumatic epilepsy (PTE), the volume of blood–brain barrier (BBB) disruption was significantly larger than that of non-epileptic patients. (C) A 34-year-old patient with PTE 1-month following mild traumatic brain injury (TBI). Power spectrum showing a marked increase in power at 1.125 Hz, taken from an electroencephalography recording, demonstrating abnormal slowing maximal at right frontal and temporal electrodes (inset). (D) sLORETA localising the pathological signal to the anterior parts of the right middle temporal gyrus (Brodmann area 21, left), and MRI signal enhancement indicating increased BBB permeability localised to the same region (right). * = p<0.05.

The spatial relationship between BBB disruption and abnormal cortical function was assessed by localising the cortical sources of the observed abnormal delta activity using sLORETA.8 In 7 of 10 patients with BBB disruption, sLORETA localised the source for abnormal delta to the same Brodmann area as the BBB disruption (table 1). Furthermore, of the 3 patients with PTE with normal EEG, two had no BBB disruption, whereas the third had not undergone an MRI scan. The volume of cortex localised with abnormal slow activity correlated linearly with the size of the BBB disrupted region (correlation coefficient = 0.53, p = 0.04), but not with the size of the cortical lesion (correlation coefficient = 0.34, p = 0.19, data not shown). Figure 1C–D demonstrates data from PTE patient #16 showing abnormal rhythmic slowing maximally recorded in right temporal electrodes and localised to the anterior part of the middle temporal gyrus (Brodmann area 21). BBB analysis showed focal pathological enhancement in the same region (Brodmann area 21).


Using quantitative EEG analysis and a semi-quantitative method for detecting abnormal BBB permeability following mild TBI, we found: (1) Increased EEG slowing in the delta-theta range; (2) Increased BBB permeability in 56% of patients; (3) Spatial correlation between focal enhanced BBB permeability and EEG delta activity; (4) Correlation between the size of the BBB disrupted region, but not that of the anatomical lesion, and the extent of cortical dysfunction; and, (5) Patients with PTE were more likely to show abnormal BBB permeability, and in larger cortical areas, compared with patients with non-epileptic TBI.

Consistent with earlier studies,7 9 patients with TBI commonly had abnormal EEG. Interestingly, spectral analyses did not reveal significant differences between our two groups. Abnormal EEG slowing reflects dysfunction of the cortical network and probably neuronal hypersynchronisation. Although animal models for brain injury consistently reveal electrophysiological evidence for neuronal hyperexcitability and hypersynchronicity,3 4 10 11 behavioural manifestations of seizures have rarely been reported.12 This raises the possibility that neuronal hypersynchrony may not necessarily manifest as convulsions,13 possibly depending on the cortical region involved. Similarities found in EEG analysis between our PTE and non-epileptic patients may be due to a common path—that is, similar pathological neuronal dysfunction presenting with different phenotypes. We note that all patients were referred to our outpatient clinic due to significant symptomatology. Further investigation of the differences between these groups and asymptomatic patients with TBI is necessary to elucidate this supposition.

We used sLORETA8 to calculate the distribution of the delta band within the cerebral cortex grey matter. The proximity of the cortical region with maximal delta activity to the MR-defined cortical lesion is suggestive of a causative relationship. We found no correlation between the volume of cortex with abnormal delta activity and the size of the anatomical lesion. This may reflect that the area of contusion undergoes neuronal necrosis and gliosis and hence displays no neuronal activity. However, the surrounding brain tissue remains functional and may be subjected to conditions that impact its normal function. The size and co-localisation of abnormal EEG activity and BBB disruption supports the supposition that lasting BBB disruption may be causally related to the emergence of neocortical dysfunction. An alternative hypothesis is that seizure activity leads to BBB opening. We believe that this is a less likely explanation in these cases, as increased BBB permeability was observed in patients despite apparent complete medical control of seizures, and no seizures were reported at least 24 hours before the brain scans. In addition, recent animal experiments directly demonstrated potential molecular pathways linking BBB leakage and increased neuronal excitability.3 Due to the retrospective nature of our study and the evidence given of PTE patients with lingering BBB disruption lasting up to several years, we can not exclude that both conditions exist, where early disruption leads to seizure activity, which in turn perpetuates the leakage. Prospective human studies are needed to further elucidate the temporal and spatial relationship between early breakdown of the BBB and development of delayed brain dysfunction, including epilepsy.


View Abstract


  • OT and IS contributed equally to this work

  • Competing interests: None declared.

  • Ethics approval: Ethics approval was obtained from the Soroka Medical Center Helsinki Committee.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.