- Top of page
- MATERIAL AND METHODS
Summary: Purpose: It is generally accepted that blood–brain barrier (BBB) failure occurs as a result of CNS diseases, including epilepsy. However, evidences also suggest that BBB failure may be an etiological factor contributing to the development of seizures.
Methods: We monitored the onset of seizures in patients undergoing osmotic disruption of BBB (BBBD) followed by intraarterial chemotherapy (IAC) to treat primary brain lymphomas. Procedures were performed under barbiturate anesthesia. The effect of osmotic BBBD was also evaluated in naive pigs.
Results: Focal motor seizures occurred immediately after BBBD in 25% of procedures and originated contralateral to the hemisphere of BBBD. No seizures were observed when BBB was not breached and only IAC was administered. The only predictors of seizures were positive indices of BBBD, namely elevation of serum S100β levels and computed tomography (CT) scans. In a porcine model of BBBD, identical procedures generated an identical result, and sudden behavioral and electrographic (EEG) seizures correlated with successful BBB disruption. The contribution of tumor or chemotherapy to acute seizures was therefore excluded.
Conclusion: This is the first study to correlate extent of acute BBB openings and development of seizures in humans and in a large animal model of BBB opening. Acute vascular failure is sufficient to cause seizures in the absence of CNS pathologies or chemotherapy.
Seizures and epilepsy are commonly observed in conjunction with stroke, traumatic brain injury, and central nervous system infections, all conditions known to result in compromised BBB function. A point of debate is whether the compromised integrity of the BBB may be a prodromic component of the etiology of epilepsy or if BBB failure is simply a consequence of seizures. In support of the former is the fact that BBB disruption after acute head trauma is a well-known pathologic finding in both animal and humans studies (Schmidt and Grady, 1993; Grant and Janigro, 2004). This disruption may persist for weeks to years after the injury and may colocalize with abnormal EEG activity (Korn et al., 2005).
The increased interest in osmotic opening of the BBB as a viable mechanism of increased drug delivery to the brain provides an opportunity to explore the connection between BBB disruption and seizures in a more controlled, yet “human” environment. Osmotic opening of the blood–brain barrier by intravascular infusion of a hyperosmolar bolus of mannitol is mediated by vasodilatation and shrinkage of capillary endothelial cells. The cells shrink resulting in widening of the interendothelial tight junctions to an estimated radius of 200 Å (Kroll and Neuwelt, 1998). The marked increase in BBB permeability to intravascular substances (10 to 100-fold for small molecules) following the osmotic disruption procedure is due to both increased diffusion and bulk fluid flow across the tight junctions. The permeability effect is largely reversed within minutes (Armstrong et al., 1989; Greig et al., 1990). Loss of BBB integrity by intrarterial hyperosmotic mannitol has been shown, in rodents, to rapidly lead to EEG changes consistent with epileptic seizures (Fieschi et al., 1980); spike/wave complexes were interspersed with decreased EEG voltage and persisted for several hours after the BBB disruption event.
Given these findings, it is not surprising that seizures are a primary complication of osmotic BBB disruption; seizures occur in a relatively large number of patients (13–55%). This was initially attributed to the use of meglumine iothalamate, a known epileptogenic agent used as a contrast agent for computed tomography (CT). Seizures continued to occur when BBB disruption was monitored by radionuclide scanning rather than CT, albeit with decreased frequency (Neuwelt et al., 1983a, 1983b). However, the correlation between level of BBB disruption and probability of seizure events has not yet been studied, nor is it clear how BBB disruption and seizure events correlate temporally.
The main goal of our study was to investigate the temporal and quantitative correlation between intraarterial BBBD with mannitol and the development of seizures in humans and in a large animal model of osmotic BBB opening. In particular, we wished to test the hypothesis that increased levels of BBB disruption are more likely to result in seizures compared to attempts leading to modest opening of the BBB. The degree of opening was quantified radiologically by contrast CT scans and by serum analysis of S100β, a serum marker of blood–brain barrier integrity (Marchi et al., 2003a,2003b, 2004; Fazio et al., 2004; Vogelbaum et al., 2005).
- Top of page
- MATERIAL AND METHODS
The patient data relevant to this study are shown in Tables 1 and 2. Fig. 1A describes the timing of mannitol, methotrexate arterial infusion, blood sampling for S100β measurement, CT scan, and probability of seizure onset (red-to-white triangle). Note that S100β serum levels were always assessed immediately after BBBD with mannitol (5′, dotted box in Fig. 1A) and before the onset on focal motor seizures either in human and pig. We first evaluated the occurrence of behaviorally detectable motor seizures when patients received intraarterial chemotherapy without BBB disruption with mannitol. Previous studies (Kapural et al., 2002; Marchi et al., 2003a, 2003b) have shown that intraarterial chemotherapy does not per se cause any significant change in blood–brain barrier integrity. In the present study, patients undergoing intraarterial chemotherapy alone without BBB disruption never experienced motor seizures (n = 5 patients for a total of 14 IAC procedures). In contrast, 25% of 102 procedures in eight patients where intraarterial chemotherapy was administered after BBB disruption resulted in motor seizures. Note that five patients, who were initially treated with IAC without BBBD during 14 cycles, later received full BBB disruption treatments. Thus, IAC without BBBD did not cause seizures in the same patients in whom IAC with BBBD did.
When timing of seizure occurrence was analyzed, we found that seizures occurred exclusively within the time frame of the procedure, i.e., minutes and not hours after the administration of mannitol and disruption of the BBB. Occasionally, patients who had seized during the procedure continued to have additional periprocedural seizures (for up to 6 h postprocedure); however, seizures occurred most commonly prior to administration of either chemotherapic agents or contrast media (Fig. 1A). Only in two procedures seizures were detected during chemotherapy after BBB disruption. There was no association between occurrence of seizures and radiologic contrast infusion, since this was performed well after the occurrence of motor seizures (Fig. 1A). These results suggested a temporal and perhaps causal relationship between BBB disruption and onset of motor seizures. These data also excluded a significant contribution of chemotherapic agents or contrast media to the observed seizure behavior.
We examined the possibility that more widespread BBB opening were associated with occurrence of seizure. In particular, we assessed the degree of BBB disruption by CT scan and measured levels of S100β increase in serum. S100β has been widely used as a surrogate serum marker of BBB integrity (Mussack et al., 2002; Jaranyi et al., 2003; Sendrowski et al., 2004; Vogelbaum et al., 2005). S100β increases sharply in blood immediately after a successful BBBD procedure (Kapural et al., 2002; Marchi et al., 2003a) and is now a recognized alternative to contrast enhanced radiological exams for BBB integrity (Vogelbaum et al., 2005). The data in Fig. 1B show the results from eight patients where serum samples to measure the serum BBB indicator S100β were obtained immediately prior and immediately after BBBD (n = 97 procedures; serum S100β data from 5 BBBD were not available). The differential serum S100βpost-BBBD-S100βpre-BBBD are indicators of blood–brain barrier leakage (Kapural et al., 2002; Marchi et al., 2003a, 2003b; Vogelbaum et al., 2005). Note that S100β was measured in blood before the onset of seizures, thus indicating goodness of the osmotic opening of the BBB and not BBB disruption induced by seizures.
In the same patients, BBBD was assessed by intraoperative CT scans (Roman-Goldstein et al., 1994a) taken approximately 30–120 min (Fig. 1A) after mannitol injection (Fig. 1C). The qualitative nil, fair, good, and excellent radiological scores were transformed into numeric values (1–4) for clarity and statistical analysis. Thus, we used a dual approach to uncover a possible link between the degree of BBBD and propensity toward the development of seizures. Note that regardless of the approach used (S100β or CT), seizures occurred preferentially during procedures associated with a successful blood–brain barrier disruption.
A direct cause–effect correlation between BBBD and motor seizures can be ascribed to disruption of brain homeostasis in proximity of the vessels where the BBB was breached. Thus, one expects that most seizures will originate contralateral to the site of BBBD, and that focal motor seizures would be more predominant during procedures affecting motor cortex (i.e., intracarotid application of mannitol) than in those where the vertebral circulation was used to deliver the osmotic agent. Our data show that this was indeed the case (Fig. 1D). BBB disruption by mannitol injection in the anterior circulation had a significantly higher probability of causing motor seizures. Furthermore, acute neurologic changes following intracarotid application almost invariably manifested as motor seizures occurring contralateral to the site of injection (Fig. 1D). It is possible that electrographic seizures arising from areas outside the motor cortex (e.g., occipital cortex) may have been occurring after disruption of the vertebral circulation.
In one patient who underwent 11 BBBD, a seizure occurred only when the radiologic index was 2 or more, suggesting that even within the same course of treatment, seizures are promoted by successful BBB disruption. In the same patient, a BBBD with a good opening (radiological score of 3, and S100β increase immediately after BBBD of 0.507 ng/ml) was sufficient to cause a seizure, while the same procedure with marginal success in disruption of the BBB administered 24 h later did not (S100β increased by 0.014 ng/ml, which is an indication of no disruption (Marchi et al., 2004)). Since the tumor burden in this patient was obviously the same at 24-h interval, we concluded that size of the tumor was not a predictor or determinant of BBBD. This was also investigated in the whole cohort of patients. There was no association between tumor size or the number of procedures undergone since the beginning of treatment and development of seizures (Fig. 2), demonstrating a lack of correlation between tumor burden and effects of BBBD on brain excitability. Fig. 2B shows the correlation between volumetric tumor sizes determined by MRI 24 h before the first BBBD procedure and the total number of seizures that these patients experienced during the entire duration of the treatment. Fig. 2C summarizes the lack of correlation between volume of tumor(s) determined by MRI the day before a given BBBD procedure and the probability of developing seizures during that particular treatment session. Note the lack of positive correlation between tumor size and propensity toward the development of motor seizures.
Figure 2. Lack of correlation between seizure occurrence (indicated by filled bars in A), tumor size or site and treatment cycle. (A) Seizure occurrence in six patients where volumetric tumor analysis was performed at each treatment episode. Each treatment refers to two subsequent BBBD at 24-h interval (see Methods). Thus, two seizure episodes may have occurred during the same treatment. The vertical lines refer to sequential treatments indicated by the numbers (1–12). The numbers at the left are patient ID's as per Tables 1 and 2. The numbers below each graph represent the MRI volume of the tumor at the time indicated. Tumor location is schematically shown in the drawing to the right. When more than one tumor site was present, two color-coded symbols are used to match their location shown to the right with their size. (B) Lack of correlation between tumor size at beginning of first treatment and total seizure numbers during the whole treatment period. (C) Cumulative tumor burden measured in patients at time of BBBD leading to seizures or not. Tumor size does not correlate with occurrence of seizures (p = 0.6). In fact, on average, smaller tumor size was present at the time of seizure occurrence further ruling out a contribution of the tumor to epileptogenicity.
Download figure to PowerPoint
As shown in Table 1, three out of eight patients (numbers 1, 2, and 8) were not prophylactically treated with antiepileptic drugs while the other five (numbers 3–7) received daily phenytoin or carbamazepine to reduce the risk of tumor-related seizures. To our knowledge, none of these patients ever experienced seizures before enrollment in the BBBD program. Interestingly, BBBD-triggered seizures occurred with equal probability in both groups, suggesting no significant influence of previous AED regimen.
While the data so far presented were all pointing to acute blood–brain barrier failure as a trigger of motor seizures, our intraoperative experimental design prevented us from directly measuring electrographic seizures in these patients. This was primarily due to logistic issues, such as removal of EEG wires during CT scanning, etc. Furthermore, while the results in Fig. 2 clearly show no correlation between tumor presence or size and seizures, we could not exclude some effect of preexisting lymphoma or ongoing peritumoral infiltration processes in determining seizures. Finally, one may object that while seizures were more often seen before MTX injection, a persistent long-term effect of chemotherapy could not be ruled out. In other words, it is unlikely but possible that previous chemotherapic exposures were responsible for lingering effects leading to seizures. To address all these issues, we used an animal model of acute BBB disruption based on an identical procedure performed by the same medical team on naive adult pigs. In these experiments, we were able to isolate BBBD from other variables (tumor, chemotherapy) and also perform EEG analysis during and after the BBBD procedure. The results of these experiments are shown in Figs. 3–5.
Figure 4. EEG correlates of acute blood–brain barrier disruption: a widespread increase in BBB permeability leads to seizures. (A) EEG recordings revealed a sharp increase in activity after BBBD in this animal. The BBBD was performed on the left hemisphere where activity was predominant. Motor seizures predominated on the right side. The traces in (B) are magnified segments as indicated by the dashed boxes in A. The arrows point to the EEG slowing that followed mannitol infusion (see also Fig. 4). Filters used for viewing during collection were 0.08 and 300 Hz (low pass and high pass, respectively); sampling rate was 1000 Hz. (C1–C2) Serum S100β increased occurred immediately post-BBBD, as assessed by western blot. Morphological demonstration of successful hemispheric blood–brain barrier disruption by Evans blue staining. Note the ubiquitous leakage in the left hemisphere and the absence of extravasation in the right. w.m: white matter; DG: dentate gyrus; fim: fimbria.
Download figure to PowerPoint
Figure 5. Histological analysis Evans blue extravasation. (A) Low-power micrographs showing the widespread leakage of the albumin-Evans blue complex (red signal) in the disrupted (BBBD) hemisphere compared to non-BBBD. The corresponding image was obtained by nuclear staining with DAPI to illustrate the relationship between serum leakage and anatomical structures. DG-dentate gyrus; s.r., stratum radiatum. (B) Higher-power images demonstrating selective leakage of the dentate gyrus in neighboring sections. (C1–C2) In the regions of Evans blue extravasation, neuronal uptake (C1) was frequently observed. Note that the “filling” of the cell extends to both basal and apical dendrites of these CA1 pyramidal cells. C2 shows vascular profiles and surrounding leakage.
Download figure to PowerPoint
A total of three blood–brain barrier disruptions were performed on two animals. Identical procedures and personnel were used. EEG electrodes were positioned on the animal's skull before BBBD as indicated in the insets. Fig. 3 shows the result of an experiment where the success of the BBBD was evaluated by measurements of S100β immediatelly before and after mannitol infusion and also by extravasation of Evans blue. The first procedure was performed on the right hemisphere, while the second was performed on the left. There was no Evans blue extravasation following the first procedure as judged by postmortem inspection of the hemisphere. Consistent with this finding was the fact that S100β levels did not change after mannitol injection (data not shown). Based on these considerations, the “goodness of opening” was deemed to be nil, according to the scale used for patient evaluation. During and after this first procedure, there were no significant EEG changes suggestive of cortical synchronization or seizure-like activity. However, there were significant EEG changes seconds after mannitol injection regardless of the site of injection (see arrows in EEG tracing in Fig. 3A). The first change was a bilateral “flattening” of EEG signals. This was not due to a filtering artifact, since reduction of EEG signal was similarly observed on unfiltered traces (see, e.g., Fig. 4). The mechanism of this phenomenon is unknown.
The second procedure was more successful and small, patchy leaky spots of Evans blue were observed in gray and white matter regions of the interested hemisphere (BBBD2 in Fig. 3D). After BBBD2, the only changes observed consisted of spike-wave complexes that were seen in both left (disrupted) and right (undisrupted) hemispheres (arrows in Fig. 3B). These changes were not accompanied by any significant behavioral seizure. Fig. 3C depicts the radiological appearance of the cerebral vasculature and the positioning of electrodes.
In a subsequent trial on a different animal, an excellent hemispheric (left) blood–brain barrier disruption was obtained after a single mannitol infusion (Fig. 4). This was evident by inspection of postmortem gross brain anatomy (Fig. 4C1) and by comparing serum S100β changes after the BBBD procedure (Fig. 4C2). The hippocampus and cortex of the disrupted hemisphere were stained with Evans blue, while the white matter appeared largely unstained. Electrophysiologically, the early EEG flattening was identical to the changes seen during other, less successful procedures (e.g., Fig. 3). Immediately after EEG suppression, high frequency and high amplitude signals appeared ispilateral to the disrupted hemisphere (Fig. 4A–B). These rapidly spread to the contralateral hemisphere. During this time, the animal experienced obvious motor seizures characterized by head elevation and body/limb extension.
As a complement to the S100β analysis as means of measurement of the level of BBBD, vascular extravasation of Evans blue was evaluated by fluorescent microscopy. The red signal shown in Fig. 5 depicts the vascular and parenchymal staining of autofluorescent Evans blue bound to albumin. Note that albumin extravasation in the left (BBBD) hemisphere was much more prominent compared to the right. Interestingly, albumin cellular uptake was seen in neurons.