Brain abscess – A rare confounding factor for diagnosis of post-traumatic epilepsy after lateral fluid-percussion injury

Objective: To assess the prevalence of brain abscesses as a confounding factor for the diagnosis of post-traumatic epilepsy (PTE) in a rat model of lateral fluid-percussion-induced (FPI) traumatic brain injury (TBI). Methods: This retrospective study included 583 rats from 3 study cohorts collected over 2009 – 2022 in a single laboratory. The rats had undergone sham-operation or TBI using lateral FPI. Rats were implanted with epidural and/or intracerebral electrodes for electroencephalogram recordings. Brains were processed for histology to screen for abscess(es). In abscess cases, (a) unfolded cortical maps were constructed to assess the cortical location and area of the abscess, (b) the abscess tissue was Gram stained to determine the presence of gram-positive and gram-negative bacteria, and (c) immunostaining was performed to detect infiltrating neutrophils, T-lymphocytes, and glial cells as tissue biomarkers of inflammation. In vivo and/or ex vivo magnetic resonance images available from a subcohort of animals were reviewed to evaluate the presence of abscesses. Plasma samples available from a subcohort of rats were used for enzyme-linked immunosorbent assays to determine the levels of lipopolysaccharide (LPS) as a circulating biomarker for gram-negative bacteria. Results: Brain abscesses were detected in 2.6% (15/583) of the rats (6 sham, 9 TBI). In histology, brain abscesses were characterized as vascularized encapsulated lesions filled with neutrophils and surrounded by microglia/ macrophages and astrocytes. The abscesses were mainly located under the screw electrodes, support screws, or craniectomy. Epilepsy was diagnosed in 60% (9/15) of rats with an abscess (4 sham, 5 TBI). Of these, 67% (6/9) had seizure clusters. The average seizure frequency in abscess cases was 0.436 ± 0.281 seizures/d. Plasma LPS levels were comparable between rats with and without abscesses (p > 0.05). Significance: Although rare, a brain abscess is a potential confounding factor for epilepsy diagnosis in animal models of structural epilepsies following brain surgery and electrode implantation, particularly if seizures occur in sham-operated experimental controls and/or in clusters.


Introduction
Brain abscess is a challenging clinical problem that can be caused by bacteria, mycobacteria, fungi or parasites (Brouwer et al., 2014).In humans, the incidence rate of brain abscess is low but increasing with estimated to range from 0.2 to 2.9 per 100,000 (Bodilsen et al., 2020a;Iro et al., 2023).In bacterial cases, brain abscesses develop in response to a parenchymal infection with pyogenic bacteria, beginning as a localized area of cerebritis and evolving into a suppurative lesion surrounded by a well-vascularized fibrotic capsule (Britt et al., 1981;Kielian, 2004).The early stage of infection is characterized by increased blood vessel permeability without angiogenesis.If untreated, this process progresses to an immature capsular stage within 1-2 weeks, and then to a brain abscess (Flaris and Hickey, 1992;Patel and Clifford, 2014).In Staphylococcus aureus-infected rats, late-stage abscesses are characterized by a necrotic core with infiltrated neutrophils, a fibrotic scar, and glial activation in the surrounding tissue (Flaris and Hickey, 1992;Zheng et al., 2019).Magnetic resonance imaging (MRI) is widely used to diagnose abscesses in humans and animals, showing rim-like enhancement (Erdogan et al., 2005;Costanzo et al., 2011;Liu et al., 2018).
Our interest in brain abscesses was kindled by observations of unprovoked seizures in a few craniectomized rats serving as experimental controls in our studies of post-traumatic epilepsy (PTE).As the diagnosis of PTE requires the exclusion of other epileptogenic factors, we systematically analyzed a total of 583 rats, including 107 shamoperated controls and 476 rats with traumatic brain injury (TBI) that had undergone both craniectomy and electrode-implantation-related surgeries.The cases were generated in 3 large animal cohorts processed for histology after long-term video electroencephalography (EEG) during the years 2009-2022(CURE cohort: Nissinen et al., 2017;EPITARGET: Lapinlampi et al., 2020;EpiBioS4Rx: Ndode-Ekane et al., 2022).Histologic sections covering the entire cerebrum were available for all 583 cases.In vivo and/or ex vivo MRIs were available for 327 cases (EPITARGET, EpiBioS4Rx) for assessing the abscess development relative to the injury and electrode implantation surgeries.Although rare, our observations indicate that brain abscesses are a potential confounding factor for a diagnosis of epilepsy in animal models of structural epilepsies that undergo brain surgery and electrode implantation.
All animal procedures were approved by the Animal Ethics Committee of the Provincial Government of Southern Finland and performed in accordance with the guidelines of the European Community Council Directives 2010/63/EU.

Induction of lateral fluid-percussion injury (FPI)
In all 3 cohorts, TBI was induced by the same technician using lateral FPI according to Kharatishvili et al. (2006).The original publications were referred to for detailed descriptions of the procedures (CURE - Nissinen et al., 2017;EPITARGET -Lapinlampi et al., 2020;EpiBioS4Rx -Ekolle Ndode-Ekane et al., 2019).No prophylactic antibiotics were used.Briefly, rats were anesthetized and placed in a stereotaxic frame.A 5-mm craniectomy was created over the left cortex with a hand-held trephine midway between lambda and bregma, and midway between the sagittal suture and temporal ridge (craniectomy center: AP -4.5 mm; ML 2.5 mm; rat brain atlas Paxinos and Watson, 2007).The intactness of the dura was confirmed visually.A plastic female Luer-lock connector made from an 18 G needle hub was placed into the craniectomy and its edges were sealed with tissue glue (3 M Vetbond, 3 M Deutschland GmbH, Germany).The connector was stabilized to the skull with a screw (1 mm, #BN82213, Bossard) placed on the left side, rostral to bregma, and surrounded by dental cement (Selectaplus, DeguDent GmbH, Germany).The rat was then removed from the stereotaxic frame and immediately connected to the FPI device (AmScien Instruments, Richmond, Virginia, USA, model 302) equipped with a straight tip.The pressure level was adjusted to produce severe TBI (expected acute post-impact mortality 25-30%).Sham-operated experimental controls underwent all surgical procedures, including craniectomy, but no pressure pulse was induced.

Postoperative monitoring
After the operation, powdered pellets were provided as supplementary food for 1-3 days or as needed.Supplementary volume correction was delivered subcutaneously using 0.9% NaCl (20 ml/24 h) for 1-3 days or longer if needed.Acute (<48 h) and follow-up (>48 h) mortalities were recorded.

EPITARGET cohort.
Two cortical screw electrodes (E636/20) were placed over the ipsilateral cortex (rostral and caudal to craniectomy) and another was placed on the contralateral side (corresponding to the center of the craniectomy) at 5 months post-TBI (see details in Lapinlampi et al., 2020).
In all cohorts, 2 electrodes were implanted into the skull bilaterally over the cerebellum to serve as a reference or a ground electrode.

Video-EEG monitoring
2.4.2.1.CURE cohort.The first continuous (24/7) video-EEG monitoring, lasting 1 week, was started at 9 weeks post-TBI.The second monitoring period started at 12 weeks post-TBI, and lasted 2 weeks (Nissinen et al., 2017).The EEG was acquired using the Nervus EEG (ver.5.71) recording system connected to an M40 (Taugagreining, Iceland) or Oxford (Medical Systems Division, UK) amplifier and filtered (high-pass filter 0.3 Hz cut-off, low-pass 100 Hz).The behavior of the rats was recorded with an infrared-sensitive WV-BP330/GE video camera (Panasonic, Kadoma, Japan) positioned above the cages.EX12LED-3BD-9 W infrared light (Bosch, Canada) was used at night to allow for continuous 24 h/d video monitoring.

EPITARGET cohort.
A 1-month-long continuous video-EEG monitoring was started 5 months after TBI (Lapinlampi et al., 2020).The EEG was recorded using the Nervus EEG recording system as described above.EEG was acquired at 2 kHz with high-pass filter of 0.5 Hz.No notch or low-pass filters were used.A WFL-II/LED15 W infrared light (Videor Technical, GmbH, Nuremberg, Germany) was used during the lights-off period.
In all cohorts, the rats were single-housed for the entire study duration in plexiglass cages and connected to an amplifier via cables, allowing for free movement.

Video-EEG analysis
In the CURE and EPITARGET cohorts, the EEG files were visually analyzed on a computer screen.In the EpiBioS4Rx cohort, each video-EEG raw data file was first imported to Spike2 (version 9, CED, UK).The occurrence of unprovoked seizures was then annotated using a seizure detection algorithm (Andrade et al., 2018), followed by a visual review of positive hits to confirm the true positive findings.All video-EEG analyses were performed in a blinded manner without knowledge of the treatment group.
EEG seizures were defined as high-amplitude rhythmic discharges that clearly represented a new pattern of tracing (repetitive spikes, spike-and-wave discharges and slow waves, frequency, and amplitude modulation) that lasted at least 10 s.If an electrographic seizure was observed, its behavioral severity was scored based on Racine's scale (Racine, 1972).Rats were diagnosed with epilepsy if one unprovoked or handling-related seizure occurred.LPS levels 2.5.1. Blood sampling 2.5.1.1. CURE cohort.No blood sampling was performed.

CURE cohort
No MRI was performed in the CURE cohort.

EpiBioS4Rx cohort
In the MRI sub-cohort, in vivo MRI imaging was performed on D2, D9, D30, and at 5 months post-TBI.In addition, all animals in the Epi-BioS4Rx cohort (both MRI and EEG sub-cohorts) were ex vivo imaged after perfusion for histology (see below).Details of the in vivo and ex vivo MRI protocols are described in Immonen et al. (2019).

Perfusion
The endpoint in the CURE cohort was 14 weeks (i.e., 14 weeks after craniectomy and 5 weeks after electrode implantation surgery), that in the EPITARGET cohort was 6 months (i.e., 6 months after craniectomy and 5 weeks after electrode implantation surgery), and that in the Epi-BioS4Rx cohort was 7 months (i.e., 7 months after craniectomy and months [EEG sub-cohort] or 5 weeks [MRI sub-cohort] after electrode implantation surgery) post-TBI or sham-operation.
Frozen coronal sections of the brain were cut (30-µm thick, 1-in-5 series) using a sliding microtome.The first series of sections was stored in 10% formalin at room temperature (RT) and used for thionin staining.Other series of sections were collected into tissue collection solution (30% ethylene glycol, 25% glycerol in 0.05 M PB) and stored at − 20 • C until processed.

Nissl staining
The first series of sections was stained with thionin, cleared in xylene, and cover-slipped using Depex® (BDH Chemical, Poole, UK) as a mounting medium.Seizure frequency was calculated as the number of seizures/number of monitoring days.Abbreviations: AP, anteroposterior; CURE, Citizens United for Research in Epilepsy; d, day; EpiBioS4Rx, the Epilepsy Bioinformatics Study for Antiepileptogenic Therapy; Sham-, sham-operated rat with epilepsy; Sham+ , a sham-operated rat without; TBI, traumatic brain injury; TBI-, a rat with TBI without epilepsy; TBI+ , a TBI rat with epilepsy;* , no area measurement was done.

Unfolded cortical maps
Unfolded cortical maps were prepared from the CURE cohort to quantify the area of any abscesses (mm 2 ) as described by Ekolle Ndode-Ekane and coworkers (2017).Briefly, an unfolded cortical map was generated from each brain and adjusted on a template prepared using the rat brain atlas of Paxinos and Watson (2007).The template comprised 26 coronal plates (bregma: +3.0 to − 9.24).Thionin-stained sections containing the abscess were then photographed using a computer-operated microscope (Leica, Germany) and digital camera (Nikon DXM1200F, Japan).The images were uploaded into the Image-J®-software as TIFF files.Three measurements were obtained from the cortical surface: (1) length from the rhinal fissure to the lateral edge of the lesion, (2) length between the ventral and dorsal edges of the lesion, and (3) length from the dorsal edge of the lesion to the medial reference point.Each section was adjusted for the shrinkage that occurs during tissue processing.Finally, each measurement was transferred to the template, and the unfolded maps of were drawn for each animal.

Gram staining
Gram staining was done to determine the presence of gram-positive and gram-negative bacteria in the abscess tissue.Two coronal sections per case covering the abscess center (n = 15, Table 1) were mounted onto glass and left to dry.First, the sections were dipped in ddH 2 O and then stained in a solution containing primary stain crystal violet 1% aqueous (3 parts, V5265, Sigma-Aldrich, Germany) and sodium bicarbonate 5% aqueous (1 part) for 1 min.The sections were then rinsed in ddH 2 O.The sections were dipped in mordant Gram's iodine stain (90107, Sigma-Aldrich) for 1 min and then rinsed with ddH 2 O. Next, decolorization was done with acetone-absolute alcohol solution for 2 min.The sections were counterstained with Gram's fuchsin working solution (87794, Sigma-Aldrich) for 1 min and rinsed with ddH 2 O. Differentiation was achieved by dipping the sections in acetone and transferring them to 0.1% picric acid in acetone solution (19553, Electron Microscopy Sciences, Hatfield, PA, USA) until a yellowish-pink color appeared (4-5 min).The sections were rinsed with acetone and transferred to acetone-xylene solution.Finally, the sections were cleared with xylene and mounted using DePex®.Gram-positive bacteria stain blue, gram-negative bacteria stain pink, and other tissues appear yellow.
The free-floating sections were rinsed 3 times in 0.02 M KPBS (10 min each) and endogenous peroxidase was removed by incubating the sections with 1% H 2 O 2 in 0.02 M KPBS for 15 min.The sections were washed (6 times, 5 min each), and non-specific binding was blocked with a solution containing 10% normal goat serum (NGS) and 0.5% Triton-X-100 in 0.02 M KPBS for 2 h at RT.The sections were then incubated in a primary antibody solution comprising primary antibody at the indicated dilution, 1% NGS, and 0.5% Triton X-100 in 0.02 M KPBS for 2-3 nights at 4 • C. The sections were washed (3 times, 10 min each) with 2% NGS in 0.02 M KPBS.The sections were then incubated in a secondary antibody solution comprising secondary antibody at the indicated dilution, 1% NGS, and 0.5% Triton X-100 in 0.02 M KPBS for 2 h at RT. Next, the sections were washed (3 times, 10 min each) and incubated with avidin-biotin complex for 2 h at RT. Finally, after washing, the staining was visualized by incubating the sections in 0.1% 3 ´,3 ´-diaminobenzidine (Pierce Chemicals, Rockford, IL) solution containing 0.04% H 2 O 2 in 0.02 M KPBS.The sections were washed once with 0.02 M KPBS and twice in 0.1 M PB, mounted on gelatin-coated slides, and covered with DePex®.

Statistical analysis
The data were analyzed using SPSS for Windows (version 25.0).The Shapiro-Wilks normality test was used to evaluate the distribution of the data.Fisher's exact test was used to analyze the difference in the prevalence of abscesses between cohorts.The difference between the groups at a given time point was analyzed by the Kruskal-Wallis test followed by the Mann-Whitney U-test as a post hoc test.Correlations between an abscessed cortical area and the presence of epilepsy and seizure frequency were analyzed with Spearman's rank correlation coefficient.A pvalue less than 0.05 was considered significant.Data are presented as whisker plots from minimum to maximum or mean ± standard deviation (SD).
Unfolded cortical maps revealed variable areas of the cortex occupied by the abscess, ranging in area from 0.06-18.93mm 2 .The abscesses tended to locate under either the screw electrode, support screw (withdrawn right after surgery), or craniectomy.Abscesses were located in the somatosensory, motor, parietal, and visual cortices (Table 1, Fig. 2D-E).

MPO
Immunohistochemistry for MPO, which marks mainly neutrophil granulocytes, indicated neutrophils infiltrating the abscess core and subarachnoid space.No immunostaining was observed outside the capsulated abscess lesion (Fig. 1E-F).

CD68
No microglia/macrophage (CD68) staining was detected in the abscesses although scattered positive staining was observed in the surrounding brain parenchyma.

GFAP
No astrocyte (GFAP) staining was observed in the abscess core or its close vicinity (Fig. 1D).Reactive astrogliosis, however, was detected outside the fibrous scar (Fig. 1D outer zone).

CD3 +
No T-cells (CD3 +) were observed in the fibrous wall of the abscess near the core or its vicinity.CD68, GFAP, and CD3 immunostaining were strongly accompanied by dark background staining (IgG) at the fibrous wall of the abscess, indicating a leaky blood-brain barrier.

MRI analysis
On T2 weighted images, abscesses had a hyperintense core surrounded by a hypointense rim (Figs.1I-K and 2B).In 2 rats in the EpiBioS4Rx MRI cohort with in vivo imaging on D2, D30, and D150 after impact (i.e., before electrode implantation), abscesses were not detectable.Abscesses were detected, however, in the ex vivo MRI image 1 month later (i.e., after electrode implantation followed by a 1-monthlong video-EEG recording) and validated in histology (Figs. 1 and 2B).
In 3 cases, ex vivo MRI scans suggested an abscess-like brain lesion.Histology, however, did not confirm the presence of an abscess.Rather, the lesion appeared to result from compression of the brain by a dorsally located screw electrode with meningeal inflammation (Fig. 3).
Supplementary material related to this article can be found online at doi:10.1016/j.eplepsyres.2024.107301.

Plasma quality
The mean A414 nm was 0.16 ± 0.04 (median 0.16, range 0.08 -0.32).Hemoglobin levels did not affect the plasma LPS concentrations (data not shown).

Plasma LPS levels
Data are summarized in Fig. 5.

Discussion
The present study investigated the prevalence of brain abscesses as a possible confounding factor for diagnosing PTE after lateral FPI.The idea for the study was kindled by observations of unprovoked seizure clusters in some sham-operated animals in the CURE cohort that were associated with the presence of a brain abscess in histologic sections leading to their exclusion from the study (Nissinen et al., 2017).Here, we systematically investigated the brain histology of 583 rats available from the CURE, EPITARGET and EpiBioS4Rx cohorts generated in our laboratory over 2009-2022.Our data showed that although brain abscesses are a rare complication after craniectomy and/or electrode implantation, seizures in sham-operated animals or the occurrence of seizures in clusters warrants exclusion of these animals due to a possible brain abscess as an etiology for epilepsy after TBI or other experimentally-induced etiology resulting from epidural or brain surgery.

Brain abscesses are rare after craniectomy and electrode implantation
In a cohort of 583 animals, histologic analysis revealed single or multiple abscesses in 2.6% (15/583) of the cases.Clinical studies suggest Fig. 4. Characteristics of epilepsy in abscess cases.(A) In the CURE cohort, 64% (7/11) of the rats with abscesses had epilepsy.The number of electrographic seizures varied from 1-18/monitoring period (21 d).In the EPITARGET cohort, 1/1 rat with an abscess had epilepsy.Seizure frequency was 23 seizures/monitoring period (30 d).In the EpiBioS4Rx cohort, 1/3 rats with abscesses had epilepsy.Seizure frequency was 0.022 (3 seizures/monitoring period) (137 d).(B) In the CURE cohort, the seizure duration varied between 14-184 s.In the EPITARGET cohort, the seizure duration varied between 31-119 s.In the EpiBioS4Rx cohort, the seizure duration varied between 54-139 s (C) In the CURE cohort, seizure-related behavioral manifestations varied between 1-5 on the Racine scale, in the EPITARGET cohort between 1-3, and in the EpiBioS4Rx cohort between 4-5.Data are presented as whisker plots with mean and minimum to maximum.Orange-filled dots indicate sham-operated animals with brain abscesses.Black-filled dots indicate TBI-induced rats with brain abscesses.that the location of the brain abscess formation is closely associated with its source (Patel and Clifford, 2014).Our animal cohort had several possible infection routes for abscess formation.First, all 583 animals underwent a circular 5-mm diameter craniectomy, with the dura left intact.Intactness of the dura after craniectomy as well as after FPI was carefully confirmed, making direct pathogen access to the intracranial cavity at the craniectomy site unlikely.In 4 cases, however, the abscess location matched that of the craniectomy.It is possible that the brain parenchyma had become accessible to bacteria via later penetration of the dura related to movement of the electrode headset covering the craniectomy.Unfortunately, we did not have MRI follow-up available in these cases.At a more chronic 5-6 month time point, this is unlikely, however, due to bone re-growth refilling the craniectomy hole.In humans, a small study series reported abscess formation as a rare (<6%) complication after decompressive craniectomy, which is a substantially more invasive procedure than craniectomy in the lateral FPI model (Albanèse et al., 2003;Callovini et al., 2021).
Another possible etiology for abscess formation relates to electrode implantation.All animals underwent surgery under aseptic conditions for implantation of epidural (CURE, EPITARGET) or both epidural and intracerebral EEG recording electrodes (EpiBioS4Rx).Rats with electrode headsets were followed for up to 7 months.This included prolonged video-EEG monitoring to diagnose PTE, requiring occasional disconnection/re-connection of the animals to the recording system.Interestingly, we had 2 abscess cases that were imaged several times over the follow-up preceding the electrode implantation.The last MRI before electrode implantation at 5 months post-TBI showed no abscess.The ex vivo MRI performed 1 month after epidural or both epidural and intracerebral electrode implantation, however, showed an abscess under the epidural electrode placement that was verified in histology.It is possible that the tips of the epidurally implanted screw electrodes had moved during the follow-up and entered the subdural space, serving as infection routes.Small electrode tip-related lesions on the cortical surface were much more common than abscesses, however, and therefore, the likelihood of subdural penetration by electrode tips as an infection route remains to be explored, particularly as the abscess prevalence (1.5%) in animals with implantation of multiple intracerebral electrodes was not any greater than that in rats with implantation of epidural electrodes (3%).Also, humans with intracerebral electrode implantation can develop abscesses despite obviously higher surgical standards than in rodent operations.A recent meta-analysis reported an abscess prevalence of 0.9% after implantation of stereo-EEG recording electrodes (Mullin et al., 2016).Also, rare cases of abscess formation were reported after implantation of deep brain stimulation electrodes (Tabaja et al., 2023).
Finally, the procedures, including craniectomy and electrode implantations, were performed by the same technician under aseptic conditions and the instruments were sterilized between animals.No antibiotic prophylaxis was used to avoid the possible effects on epileptogenesis.The prevalence of abscesses decreased in the order of CURE 4.3% > EpiBioS4Rx 1.6% > EPITARGET 0.7%.Importantly, the CURE cohort generated during 2009-2012 with the highest abscess prevalence was the first large cohort performed in the laboratory.Further training in aseptic and surgical techniques and the use of sterilizing equipment may have reduced the abscess rate in later cohorts.
To summarize, craniectomy and electrode implantation for diagnostic EEG recording present the 2 most likely etiologic factors for abscess development in the lateral FPI model of PTE.

Capsulated brain abscesses tended to locate under cortical electrodes, supportive screw or craniectomy
Histopathologic examination of brain abscesses revealed vascularized encapsulated lesions surrounded by astrogliosis, indicating the late stage of capsule formation in our animal cohort (Enzmann et al., 1985;Flaris and Hickey, 1992).Although the pathogen responsible for the abscess formation remained unknown, the capsule histopathology was comparable to that generated by experimental gram-positive S. aureus-induced brain abscesses (Flaris and Hickey, 1992;Liu, 2018;Zheng, 2019).The formation of a thick astroglial scar was probably protective as GFAP 0/0 mice with S. aureus-induced brain abscess show more severe lesions, a higher central nervous system bacterial load, and a lack of bordering function (Stenzel et al., 2004).Consistent with previous studies of induced brain abscesses, we found MPO-positive infiltrated neutrophils and gram-negative bacteria in the abscess core (Flaris and Hickey, 1992;Enzmann et al., 1985;Zheng, 2019).Interestingly, Kielian et al. (2001) showed that neutrophil-depleted mice exhibited more severe brain abscesses and a higher central nervous system bacterial burden, suggesting that neutrophils play a protective role in the host response against bacterial infection.In accordance with the experimental data, a recent study in humans demonstrated that the greater the MPO level in the extracellular fluid of brain abscess pus, the greater the abscess volume (Hassel et al., 2018).
MRI is used for diagnosing brain abscesses in humans and animals (Erdogan et al., 2005;Costanzo et al., 2011, Liu, 2018).In the present study, 181/583 cases were imaged in vivo on D2, D7-D9, and D30, or months after craniectomy and at the end of the follow-up with ex vivo MRI.Consistent with previous reports of experimentally induced abscesses, abscesses showed a hypointense rim associated with iron deposits in histology.In a few cases, MRI-suspected abscesses turned out to be small inflammatory lesions, e.g., along the electrode tract with an iron deposit.This finding highlights the need for histologic analysis in MRI-suspected abscess cases.

Brain abscess formation was associated with a high prevalence of epilepsy
Although brain abscesses are a rare complication the in lateral FPI model, they are highly epileptogenic.The diagnostic video-EEG was performed 12 weeks after TBI in the CURE cohort with an epilepsy rate of 15% (Nissinen et al., 2017), 5 months after TBI in the EPITARGET cohort with an epilepsy rate of 28% (Lapinlampi et al., 2020), and months after TBI in the EpiBioS4Rx cohort with an epilepsy rate of 24% in non-abscess TBI cases (Ndode-Ekane et al., submitted).In the abscess cohort, including both the sham-operated controls and rats with lateral FPI, the epilepsy rate was 60% -more than 2-fold higher than that in rats with TBI only.If only the 9 TBI rats with abscesses were included, the epilepsy rate was 56%.Thus, the presence of a dual epileptogenic pathology (i.e., abscess and TBI) exceeded the average epilepsy prevalence of 25% after lateral FPI by more than 2-fold (Kharatishvili et al., 2006;Shultz et al., 2013;Lapinlampi et al., 2020).Interestingly, even though the animal cohort with abscesses is small, the epileptogenicity is comparable to that in humans with a brain abscess (Northcroft et al., 1957;Beller et al., 1973;Legg et al., 1973;Nicolosi et al., 1991;Nielsen et al., 1983;Koszewski, 1991;Kilpatrick, 1997;Chuang et al., 2010;Helweg-Larsen et al., 2012;Brouwer et al., 2014;Laulajainen-Hongisto et al., 2016;Lee et al., 2018;Zelano et al., 2020;Bodilsen et al., 2020Bodilsen et al., , 2023;;).
Typical of abscess-related epilepsy was the occurrence of seizures in clusters, i.e., ≥ 3 seizures/24 h as 67% of the rats had seizure clusters.Clusters were observed in sham-operated experimental controls as well as in rats with lateral FPI having an abscess.Seizure clusters are not uncommon in rats with PTE after lateral FPI as approximately 38% of the rats with PTE in the EPITARGET and 37% in the EpiBioS4Rx cohort exhibited clusters (Casillas-Espinosa et al., 2019;Manninen et al., 2020;Ndode-Ekane et al., submitted).Still, the prevalence of cluster cases in the abscess cohort was over 1.5-fold higher than that in rats with PTE induced with lateral FPI.In correlation, the average seizure frequency during the monitoring period was approximately 2-fold higher than the reported seizure frequency in rats with PTE after lateral FPI (Kharatishvili et al., 2006;Manninen et al., 2020).The mean seizure duration and behavioral severity assessed using the Racine score were comparable, however, between the abscess cohort and data available from previous studies on lateral FPI-induced epilepsy (Kharatishvili et al., 2006;Manninen et al., 2020).
Clinical studies suggest that large abscesses and abscesses with a frontal/parietal/temporal location could be associated with an increased risk of epilepsy (Koszewski et al., 1991;Brouwer et al., 2014;Lee et al., 2018;Bodilsen et al., 2020Bodilsen et al., , 2023)).Although our animal abscess cohort is small, it is interesting to note that a single abscess was found in 5 of 6 abscess cases without epilepsy and only 1 of 7 cases with multiple abscesses did not have epilepsy.We found no association between the size of the cortical area occupied by the abscess and the presence of epilepsy or seizure frequency.Also, no apparent association of epilepsy with any rostrocaudal or lobar location of the abscess was detected.
In a recent large epidemiologic study, the median time from abscess to epilepsy diagnosis was 0.76 years, i.e., approximately 9 months (Bodilsen et al., 2020).In correlation, our MRI study in rats suggests that it takes only days to a few weeks for an abscess to develop.Epileptogenesis, however, is a "moving target" and the prevalence of epilepsy depends on the timing of the EEG analysis.The EEG follow-up after electrode implantation in abscess cases without epilepsy might have been too short to detect ongoing epileptogenesis.

Further considerations -Possible pathogens and mechanisms
Although we assumed that abscess formation related to the surgical operation, the type and origin of the pathogen remains unclear.As this was a retrospective study, bacterial analysis of the pus was not possible.We did, however, perform histologic Gram staining of the fixed brain tissue, which suggested that the pathogen was gram-negative.
Next, we measured the levels of LPS, a component of the outer membrane of gram-negative bacteria, in the peripheral blood.It is used as a reporter of increased intestinal permeability and endotoxemia, and thus elevated plasma LPS levels could indicate hematologous spread of the bacteria.LPS is also used to induce peripheral infection in animals, resulting in a lowered seizure threshold to chemoconvulsant exposure (Sayyah et al., 2003;Wang et al., 2021).Interestingly, elevated LPS levels were recently proposed as potential biomarkers for PTE after lateral FPI (Mazarati et al., 2021).
Sham-operation, TBI, and abscesses increased the plasma LPS levels from 15 pg/ml to approximately 28 pg/ml.This observation suggests that even a mild TBI (i.e., craniectomy) or electrode implantation -related lesion, was associated with an elevated plasma LPS level that was not further augmented by TBI or brain abscess.Two previous studies measured plasma LPS levels after experimental TBI.In the lateral FPI model, Mazarati and colleagues (2021) reported undetectable plasma LPS levels at baseline and at 7 months post-craniectomy in sham lateral-FPI animals.In a weight-drop model with craniectomy and exposed dura, Hang et al. (2003) reported plasma LPS levels in sham-operated controls comparable to those in the present study.In both models, the post-TBI plasma LPS levels were comparable to those in the present study (Hang et al., 2003;Mazarati et al., 2021).We were unable to reproduce the elevated plasma LPS levels in TBI+ animals compared with TBI-animals as reported by Mazarati and colleagues (2021).Further studies using a larger number of animals are needed to explore the possible value of plasma LPS as a diagnostic biomarker for PTE.
Why are abscesses so highly epileptogenic?A previous clinical study demonstrated a remarkable increase in extracellular brain excitatory amino acid levels (glutamate and aspartate) in the pus aspirated from patients with an intracerebral abscess, particularly if they had seizures (Dahlberg et al., 2014).Similarly, in rats, extracellular brain tissue glutamate and aspartate levels were remarkably increased within 20 h after S. aureus injection-induced focal inflammatory striatal lesions (Hassel et al., 2014).Previous studies demonstrated that the abscess capsule does not constitute a diffusion barrier (Yamamoto et al., 1993;Lo et al., 1994), and therefore increased diffusion of excitatory amino acids from the abscess to surrounding regions remains a mechanism of ictogenesis to be explored.
Finally, we detected heavy background immunostaining in the abscess cases, indicating a leaky blood-brain barrier.Blood-brain barrier permeability and increased parenchymal albumin are associated with epilepsy both in experimental models and human epilepsy (van Vliet et al., 2007).Interestingly, serum albumin is detected in brain abscess pus in humans (Hassel et al., 2018), suggesting another potential factor contributing to ictogenesis.

Conclusions
Brain abscesses are a rare complication of the lateral FPI-induced TBI model of PTE.Due to the high epileptogenicity of a brain abscess, the possible presence of an abscess as a confounding factor for an epilepsy diagnosis should be evaluated if a sham-operated animal has seizures or if seizures occur in clusters.As brain surgery for model generation and/ or electrode implantation are commonly used in experimental epilepsy research, these data have implications beyond the models of PTE.

Fig. 1 .
Fig. 1.Abscess case #1347.(A) A photomicrograph of a rat brain with traumatic brain injury (TBI) but no epilepsy (TBI-, case #1347).The abscess (arrowhead) was located in the left hemisphere rostroparietal to the TBI lesion (asterisk).(B) A low magnification photomicrograph of a thionin-stained coronal section from the brain of the same rat.The abscess is indicated by an arrowhead.(C) A higher magnification photomicrograph of the abscess in panel B. Note the mineralization(lighter blue spots indicated with arrowheads).(D) A high-magnification photomicrograph of a glial fibrillary acid protein (GFAP)-stained section of the same abscess.Note the lack of immunopositive cells in the abscess core (arrowhead) and in the surrounding tissue (area between solid lines).Reactive astrogliosis was observed outside the scar tissue.Note also the high background staining, suggesting blood-brain-barrier leakage.(E) A photomicrograph of an abscess stained with myeloperoxidase (MPO).Note the intensive immunopositive neutrophil staining (dark brown), specifically in the abscess core.(F) A higher-magnification photomicrograph of the MPO staining in panel E, showing the morphology of a neutrophil (insert F1).(G) A Gram-stained abscess.(H) A higher magnification photomicrograph of the Gram staining in panel G.Note the gram-negative (red) granular staining in the abscess core.The yellow color indicates background staining.(I) Coronal, (J) sagittal, and (K) horizontal planes of ex vivo magnetic resonance images of the abscess (arrowhead in all panels) of the same case, showing a hyperintense core surrounded by a hypointense rim.

Fig. 2 .
Fig. 2. Representative abscess cases.(A-C) Low-magnification photomicrographs of thionin-stained coronal sections showing an abscess (arrowhead) in a shamoperated rat with epilepsy (sham+ rat #1051).(D) Coronal, (E) horizontal, and (F) sagittal planes of the ex vivo magnetic resonance images of the same rat.The abscess is indicated by an arrowhead.The white arrow in panel F indicates a trace generated by the intrahippocampal electrode.(G-I) Low magnification photomicrographs of thionin-stained coronal sections showing multiple abscesses (arrowhead in G-H, arrow in I) in a TBI rat with epilepsy (TBI+, rat #212), one in the right (G-H; G most rostral, I most caudal) and another in the left (I) hemisphere.(J) Unfolded cortical map of the left hemisphere of rat #212, indicating the location and area of the TBI-induced cortical lesion (green) and abscess (purple) under the craniectomy (center coordinates AP − 4.5 mm, ml 2.5 mm).(K) Unfolded cortical map of the right hemisphere in rat #212, indicating the location and area of 2 abscesses (purple) under a screw electrode-induced cortical lesion (yellow).

Fig. 3 .
Fig. 3. Suspected abscess cases in MRI.(A) Coronal, (B) horizontal, and (C) sagittal planes of ex vivo magnetic resonance images (MRI) showing a suspected abscess (arrowhead) case (#1231) with a hypointense rim surrounding a hyperintense core.(D-F) Low-magnification photomicrographs of thionin-stained coronal sections from the same rat (#1231) showing a trace of the electrode tip (arrowheads) but no abscess in histology.Scattered iron deposits around the cavity matched with the signal hypointensity in the MRI image (panel A). (G) A coronal view of ex vivo MRI showing another suspected abscess case (#1232).(H) A low-magnification photomicrograph of a thionin-stained coronal section from the same rat #1232, showing a cortical indentation caused by the electrode tip (box) but no abscess in histology.Scattered iron deposits around the cavity matched the signal loss area (with the hypointense rim) in the MRI image (panel G). (I) A photomicrograph showing the compression caused by the cortical electrode tip (between the arrowheads; from box in panel H).Note the pial inflammation.

Table 1
Details of each rat with a brain abscess.Of the 15 rats, 9 (60%) had epilepsy.