Characterization of Vasogenic and Cytotoxic Brain Edema Formation After Experimental Traumatic Brain Injury by Free Water Diffusion Magnetic Resonance Imaging

Abstract

Brain edema formation is a key factor for secondary tissue damage after traumatic brain injury (TBI), however, the type of brain edema and the temporal profile of edema formation are still unclear. We performed free water imaging, a bi-tensor model based diffusion MRI analysis, to characterize vasogenic brain edema (VBE) and cytotoxic edema (CBE) formation up to 7 days after experimental TBI. Male C57/Bl6 mice were subjected to controlled cortical impact (CCI) or sham surgery and investigated by MRI 4h, 1, 2, 3, 5, and 7 days thereafter (n = 8/group). We determined mean diffusivity (MD) and free water (FW) in contusion, pericontusional area, ipsi- and contralateral brain tissue. Free (i.e., non-restricted) water was interpreted as VBE, restricted water as CBE. To verify the results, VBE formation was investigated by in-vivo 2-Photon Microscopy (2-PM) 48h after surgery. We found that MD and FW values decreased for 48h within the contusion, indicating the occurrence of CBE. In pericontusional tissue, MD and FW indices were increased at all time points, suggesting the formation of VBE. This was consistent with our results obtained by 2-PM. Taken together, CBE formation occurs for 48h after trauma and is restricted to the contusion, while VBE forms in pericontusional tissue up to 7 days after TBI. Our results indicate that free water magnetic resonance imaging may represent a promising tool to investigate vasogenic and cytotoxic brain edema in the laboratory and in patients.

Introduction

With more than 27 million new cases per year, traumatic brain injury (TBI) is a major health and socioeconomic problem with severe impact on global health.^1,2 Despite multiple research efforts, no causal therapy is available for patients with TBI yet.³ Therefore there is an urgent need for the development of novel and effective therapeutic strategies.

The pathology of TBI is complex, involving both primary and a wide range of secondary injury mechanisms.^3,4 After the initial mechanical impact, primary injury events such as tissue necrosis, diffuse axonal injury, and cerebrovascular damage occur immediately. The irreversible primary injury⁵ then initiates multiple secondary processes—e.g., neuroinflammation, microvasculatory disturbances, brain edema formation, and/or cerebral ischemia—that result in further damage of the brain and subsequent neurological deficits.^4,6–9 Because of their delayed appearance, these secondary processes are, in principle, amenable to therapy, provided the underlying pathophysiology is understood properly.^9,10

Brain edema, defined as a net increase of brain water content, is still a critical contributor to secondary brain injury after TBI.^9,11,12 Brain edema is classified into vasogenic, cytotoxic, ionic (osmotic), interstitial (hydrocephalic), and hydrostatic edema.^13,14 After TBI, mainly the classic forms —vasogenic and cytotoxic brain edema (CBE)—are discussed to determine patient outcome.^9,15–17

Pathophysiologically, CBE is based on the swelling of astrocytes after the uptake of ions and causes water movement from the extracellular space into the cells.¹⁴ Brain swelling occurs secondarily, when the extracellular space is refilled with water.¹⁷ Vasogenic brain edema (VBE), by contrast, results in an influx of water into the brain across the damaged blood–brain barrier (BBB), and therefore directly causes brain swelling.^14,17

Despite the importance of brain edema for the pathophysiology of TBI, many highly relevant aspects of the pathogenesis of brain edema formation are still not fully understood. First, it is unclear which type of brain edema occurs after TBI. Some studies suggest that CBE may be the predominant form,^15,16 while other studies also emphasize the significance of VBE formation.^9,14,17

Another open point is the temporal profile of brain edema formation. While some studies suggest that brain edema develops and resolves within several hours or a day, other investigations point out that this process may last for days or even weeks.^9,17

Finally, also the spatial occurrence of the several forms of brain edema is widely discussed. Studies using water diffusion magnetic resonance imaging (MRI) suggested the dominance of CBE in contusion and penumbra after experimental TBI,^18–20 while studies using microscopic techniques in vivo demonstrated the occurrence of VBE in the traumatic penumbra.^21,22

The ideal method to overcome this knowledge gap would be a technique that is able to measure VBE and CBE simultaneously, with a satisfactory spatial resolution, in the whole brain, and longitudinally. A novel technique that should be able to separately identify CBE from VBE in certain scenarios, free water imaging (FWI), became available in recent years.^23–27

The FWI was developed originally to correct the diffusion signal for cerebrospinal fluid partial volume effects. Quantifying such partial volume, however, also characterizes water diffusion in the extracellular space in the brain.^28,29 FWI is based on diffusion-weighted MRI (DWI), a widely used MRI technique that quantifies the diffusion properties of water molecules in tissues.³⁰

The FWI explicitly models a free water compartment—that is, water molecules that are freely moving—which in typical DWI experiments are found in the extracellular space. The remaining water molecules—that is, water in the vicinity of cellular membranes—are modeled by a second compartment, termed the tissue compartment.^29,31

Thus, FWI is able to quantify the free water fraction and to estimate microscopic measures that describe the tissue compartment without the effects of extracellular free water.^32,33 Increases in free water indicated vasogenic edema, while a decrease in free water indicates a shift of water from extracellular to intracellular space because of cytotoxic edema.

In the current study, we used FWI to investigate the spatial and temporal profile of brain edema formation up to seven days after controlled cortical impact (CCI) in mice, an animal model for contusional TBI.

Methods

Animals

In the current study, we used 30 male C57Bl/6 mice with an age of 6 to 8 weeks (Charles River Laboratories, Germany). All animal experiments were approved by the Ethical Committee of the Government of Upper Bavaria. The results of the study are reported in accordance with the ARRIVE guidelines.³⁴ Animal husbandry, health screens, and hygiene management checks were performed in accordance with the guidelines and recommendations by the Federation of European Laboratory Animal Science Associations (FELASA).³⁵

Animals were randomly assigned to the different experimental groups. Experiments and data analysis were performed by an investigator blinded with respect to the experimental groups.

Experimental TBI

The CCI model was used to induce a severe traumatic contusion as described previously.^36,37 Briefly, 30 min after an injection of buprenorphine (0.1mg/kg, single shot intraperitoneally [ip]), animals were anesthetized with 5% isoflurane in a gas mixture of 30% oxygen and 70% nitrogen for 2 min. To maintain anesthesia, isoflurane application was reduced to 1.5–2% thereafter, and the body temperature was maintained at 37^°C by a feedback-controlled heating pad.

The head of the mouse was fixed in a stereotaxic frame, and a rectangular craniotomy (4 × 4 mm) was prepared posterolateral to bregma over the right parietal cortex under continuous cooling with saline, leaving the dura mater intact. Subsequently, CCI was induced by a custom-made device (L. Kopacz, University of Mainz, Germany; diameter: 3.0 mm, velocity 6 m/sec, penetration depth 0.5 mm, contact time of 150 msec).

The craniotomy was resealed with the previously removed bone flap, the skin incision was sutured, and animals were placed in an incubator warmed at 37^°C for recovery. Animals of the sham group underwent the same surgical procedure without CCI.

MRI and data analysis

Mice were anesthetized with isoflurane (5% for induction and 1–1.5% for maintenance of anesthesia in a gas mix of 30% oxygen and 70% nitrogen), and body temperature, heart rate, and respiration rate were continuously monitored. The MRI scans were performed on a 3T nanoScan^® positron emission tomography (PET)/MR scanner (Mediso, Münster, Germany), including a T1-weighted, a T2-weighted, and a diffusion tensor imaging (DTI) sequence.

The T1-weighted imaging was based on a three-dimensional (3D) gradient echo sequence (GRE) with the following parameters: field-of-view (FOV) = 30 × 30 mm, slice thickness = 0.4 mm, matrix = 256 × 256, echo-time (TE) = 3.76 msec, repetition time (TR) = 30 msec, number of excitations (NEX) = 2, flip angle = 35 degrees.

A two-dimensional fast spin echo (FSE) sequence was used for acquiring T2 images with FOV = 30 × 30 mm, slice thickness = 0.4 mm, matrix = 256 × 256, TE = 49.74 msec, TR = 6254 msec, and NEX = 8.

The DTI was based on a single-shot echo planar imaging sequence (SS-EPI). The imaging settings were as follows: FOV = 28 × 28 mm, slice thickness = 0.8 mm, matrix = 80 × 80, TE = 55.27 msec, TR = 3000 msec, and NEX = 3. Thirty directions were obtained with b-value = 1000 s/mm² and three b0 scans were distributed among these directions. An additional b0 scan was performed with reversed phase encoding gradient for correction purposes.

Each imaging session (T1, T2 plus FWI) lasted 25 min. We scanned 14 slices from bregma level 1.1 mm to bregma level −4.1 mm. Imaging was performed 4h, 24h, 48h, 72h, 5d, and 7d post-CCI or sham operation. In addition, four naïve animals were investigated (Fig. 1A).

FIG. 1.

Timeline of the experimental setup for magnetic resonance imaging (MRI) scanning and 2-photon microscope imaging. (A) Animals were subjected to controlled cortical impact (CCI) or sham operation (n = 8 per group), and MRI was performed repetitively at six time points until 7 days after injury. In addition, four naïve animals were imaged by MRI. (B) Animals were subjected to CCI or sham operation (n = 5 per group), and 2-Photon microscopy was performed repetitively every 30 min between 48h and 50h thereafter. (C) Illustration of contusion (black), penumbra (grey), cranial window and the location of the three regions of interest on the brain for 2-Photon microscope imaging.

Lesion volume, hemispheric swelling, and midline shift were determined similar to previous histological analyses³⁸ based on T2-weighted sequences by using ImageJ (https://imagej.nih.gov/ij/) as described previously.^39,40 Briefly, we selected 14 MRI slices equidistant to each other covering the lesion from frontal to rostral. The lesion area and the area of each hemisphere were quantified (A) and lesion volume and hemispheric volume (V) were then calculated using the following formula: V = d*(A1/2 + A2 + A3…+ An/2), with d being the distance between slices in millimeters.

Hemispheric swelling was calculated as the ratio of ipsilateral to contralateral hemispheric volume. Midline shift was determined -1.7 mm from bregma, which was the section showing the maximum displacement of the lateral ventricle. The distance between the outer border of the cortex and the middle of the third ventricle was measured from the ipsilateral (A) and contralateral (B) side. Midline shift was then calculated using the following formula: Length = (A-B)/2.

Diffusion measures were calculated using the MATLAB-based FWI toolbox as described previously.^25,26,41 Briefly, in addition to the mean diffusivity (MD) acquired from conventional DTI measurements, a bitensor model-based diffusion MRI analysis was applied. Here, the first tensor fits the so-called free water compartment—i.e., is an isotropic tensor with a fixed diffusion constant of water at 37°C in the brain.

The second tensor fits the FW-corrected diffusion data to obtain the so-called tissue compartment. The fractional volume of the FW compartment shows the relative contribution of FW in each voxel, ranging from 0 to 1.^25,29 An increase in FW was interpreted as VBE, while a decrease in FW was interpreted as a proxy for CBE.

By analyzing the diffusion metric maps on the ITK-SNAP software,⁴² we determined MD and FW separately in contusion, pericontusional area, ipsilateral brain tissue (without contusion and pericontusional area) and contralateral brain tissue.

The pericontusional area was defined as the area within two voxels adjacent to the visible contusion—i.e., with the increase of the contusion, the pericontusional area moved accordingly. Hence, the pericontusional area is congruent with the traumatic penumbra as long as the contusion expands—i.e., during the first 24h (Fig. 2A). Thereafter, it reflects tissue around the fully expanded contusion.

FIG. 2.

Post-traumatic structural damage and brain swelling at different time points after controlled cortical impact (CCI) or sham operation. (A) Representative T2-weighted magnetic resonance images showing lesion volume and midline shift (white dashed line) over time post-trauma (right panel), while no significant changes were visible in the sham group (left panel). Quantification of lesion volume (B), midline shift (C) and hemispheric swelling (D). Data are presented as mean ± standard deviation; n = 4 for the naïve group, n = 8 each for the sham operated and for the trauma group. Unpaired t test for parametric data and Mann-Whitney Rank Sum test for non-parametric data, *p < 0.05, **p < 0.01, and ***p < 0.0001 vs. sham.

In vivo 2-Photon microscopy

The study used 2-Photon microscopy to verify the data obtained by MRI—i.e., to verify the formation of VBE in the pericontusional area. Mice received an ip injection of 0.05 mg/kg fentanyl, 0.5 mg/kg medetomidine, and 5 mg/kg midazolam, were intubated with a custom-made tracheal tube, and ventilated in a volume-controlled mode (Minivent 845, Hugo Sachs Electronik, March-Hugstetten, Germany) with continuous end-tidal pCO₂ monitoring by microcapnometry (Capnograph 340, Hugo Sachs Elektronik). A catheter was inserted into the right femoral artery for blood pressure measurement and fluid infusion as described previously.^21,37,43 The body temperature was controlled throughout the procedure and maintained at 37°C.

Animals were head-fixed using a stereotactic frame, and a 2 mm square craniotomy was performed over the right frontoparietal cortex under continuous cooling with saline. After gentle removal of the dura mater, the surface of the brain was rinsed with saline. Subsequently, the craniotomy was sealed with a cover glass (Schott Displayglas, Jena, Germany) using dental cement (Cyano Veneer; Hager & Werken, Duisburg, Germany).

In vivo 2-photon microscopy imaging was performed as described previously^21,37,44 using a Zeiss LSM 7MP microscope equipped with a 20 × water immersion objective (Plan Apochromat, Zeiss, Germany) and a Li:Ti laser (Chameleon, Coherent) tuned to 800 nm 48–50h after TBI. To determine BBB permeability, we used a fluorescent dye (tetramethylrhodamine/TMRM-dextran) with a molecular weight of 40 kDa (TMRM, MW 40,000; Invitrogen, Darmstadt Germany, 40 mg/kg in 100 μL injected intravascularly). As described previously, a marker of this size only extravasates after BBB disruption.^21,37,45

Z-stack images were acquired in three regions of interest (ROI) with a size of 425 × 425 μm from the surface to a depth of 300 μm into the parenchyma (Fig. 1B, 1C). Baseline images were acquired from each ROI within 2 min after injection of TMRM, and then imaging was performed every 30 min for 120 min. The animals were sacrificed at the end of the experiment.

Images were analyzed using the ImageJ analysis software (https://imagej.nih.gov/ij/) as described previously.^21,37 The superficial 50 μm of the Z-stacks were excluded from the analysis to avoid potential artifacts caused by the cranial window preparation. To analyze TMRM fluorescence intensity in the parenchyma, the threshold of the images was adjusted to eliminate both the background signal (very low intensity) and the intravascular signal (high intensity) in a standardized manner applied to all datasets. The remaining fluorescence intensity in the parenchyma was determined by ImageJ. The values were then corrected by the measurements obtained in sham operated mice to correct for bleaching.

Histological analysis

To correlate our MRI findings with histological alterations, especially white matter demyelinization, we performed Luxol^® fast blue (LFB) staining on histological sections. Animals were sacrificed in deep anesthesia by transcardial perfusion with sodium phosphate buffer (PBS; 50 mM at pH 7.4) followed by 4% paraformaldehyde (PFA) as described previously.⁴⁶ Brains were removed and stored in 4% PFA at 4^°C overnight for post-fixation and then stored in PBS at 4^°C until further analysis.

Two consecutive sets of 10 μm thick coronal sections were cut at 500 μm intervals using a cryotome (CryoStar NX70, Thermo Fisher, Waltham, MA). To visualize the lesion area and the white matter, two consecutive sets of brain sections were stained with Cresyl violet (Nissl) or LFB, respectively, and investigated using an epi-fluorescence microscope (Axio Imager.M2, Carl Zeiss, Germany). The Cresyl violet stained sections were imaged at 1 × resolution, and the lesion volume was defined as described previously.^38,44

In a second step, the white matter was determined in sections stained with LFB in three ROI: two ROI were adjacent to the lesion volume, covering the corpus callosum close to the midline and the capsula interna on the lateral side. The third ROI was placed accordingly covering the capsula interna on the contralateral side. The density of the myelin stain was quantified with the software ImageJ (https://imagej.nih.gov/ij/). For this analysis, two sections per animal were chosen from bregma level −1.0 mm to bregma level −1.5 mm.

Statistical analysis

All data were tested for normal distribution with the Shapiro Wilk test. When comparing two groups of normally distributed data, the Student t test was used. For non-normal distributed data, the Mann-Whitney Rank Sum test was used. Repeated measurements over time were normally distributed, and the two-way repeated measures analysis of variance was used, corrected by the Tukey multiple comparisons test. A statistically significant difference between groups was assumed at a p value <0.05. All calculations were performed with GraphPad Prism 9.0, and all data are given as mean ± standard deviation (SD).

Results

Physiological parameters

During MRI, respiratory rate, heart rate, and core body temperature were continuously monitored, were within the physiological range and did not differ between groups (Supplementary Fig. 1 and Table 1). During 2-photon microscopy, end-tidal pCO₂, mean arterial blood pressure, and core body temperature were continuously monitored and did not differ between groups (data not shown).

Table 1.

Core Body Temperature at the End of Magnetic Resonance Imaging

Group (n = 4–8)	Core body temperature at end of scan (°C)
Sham – 4h	38.2 ± 0.50
CCI - 4h	38.3 ± 0.22
CCI – 24 h	38.4 ± 0.87
CCI – 48 h	38.3 ± 0.39
CCI – 72 h	38.3 ± 0.21
CCI – 5d	38.5 ± 0.42
CCI – 7d	38.2 ± 0.51
Naive	38.4 ± 0.34

CCI, controlled cortical impact.

Mean ± standard deviation. The values do not differ between groups and are within the physiological range.

In addition, arterial blood gases were measured at the end of each experiment (Table 2). All values were within their respective physiological range and did not differ significantly between groups or within the groups at different time points.

Table 2.

Blood Gas Analysis After 2-Photon Microscopy

Group (n = 5)	pH	pCO2 (mm Hg)	pO2 (mm Hg)
Sham	7.34 ± 0.02	39 ± 2	116 ± 18
CCI	7.30 ± 0.04	41 ± 5	112 ± 14

CCI, controlled cortical impact.

Mean ± standard deviation. Animals are in anesthesia, intubated, and ventilated. There are no differences between groups ,and the values are within the physiological range.

Lesion volume and brain swelling after CCI by T2-weighted MRI

The contusion, hemispheric swelling, and midline shift were clearly detectable on T2-weighted MRIs, while no pathology was present in sham operated mice (Fig. 2A). The lesion volume reached a maximum of 32.7 ± 1.9 mm³ at 24h and gradually decreased thereafter (Fig. 2B, p < 0.01 vs. sham). The volume of the contralateral hemisphere was 113.2 ± 3.4 mm³ on average. Midline shift and hemispheric swelling were already present 4h after CCI, and both reached their peak between 24h and 48h after trauma (Fig. 2C, 2D, p < 0.05 and p < 0.0001 vs. sham). Seven days after TBI, no midline shift could be observed, while hemispheric volume was still elevated.

Characterization of VBE and CBE by FW MRI

The FW maps were generated from sham operated and traumatized mice (Fig. 3). Diffusion parameters were assessed by 3D analysis in four ROIs covering the contusion, the pericontusional area (the traumatic penumbra), and the ipsi- and contralateral hemisphere (Fig. 3C).

FIG. 3.

Representative images of processed magnetic resonance imaging maps obtained following trauma and sham operation. Examples of conventional mean diffusivity (MD) maps (A) and free water (FW) maps (B) after sham operation and controlled cortical impact (CCI) at 4h and 7d after trauma or sham operation (white arrows point to the contusion). After trauma, the images show a decreased MD and FW signal intensity (dark blue pixels) in the contusion at 4h post-CCI, while an increased intensity (green or light yellow pixels) is visible in the pericontusional area. At 7 days after CCI, there is still an increase in MD and FW signal intensity visible in the pericontusional area, whereas the differences within contusion are much less pronounced. (C) Schematic diagram of the regions of interest for the analysis—i.e., contusion, pericontusional area, ipsi- and contralateral hemisphere.

The MD and FW values were stable and did not show significant differences in the contralateral brain or in the ipsilateral brain outside from the contusion and pericontusional area (Fig. 4A-D).

FIG. 4.

Analysis of diffusivity measures in the contralateral hemisphere and ipsilateral without contusion and pericontusional area (A,B). Compared with sham operated or naïve animals, no significant changes are visible in mean diffusivity (MD) and free water (FW) signals of the contralateral and ipsilateral brain at all time points after trauma. Data are presented as mean ± standard deviation; n = 4 for the naïve group, n = 8 for sham and controlled cortical impact (CCI). Unpaired t test for parametric data and Mann-Whitney Rank Sum test for non-parametric data, **p < 0.01 vs. sham.

When looking for specific changes within the contusion and the pericontusional area, we observed decreased MD and FW values within the contusion and increased values in the pericontusional area already 4h after TBI (Fig. 5A, 5B). The values for MD within the contusion were significantly lower than the values in the contralateral brain until 72h after trauma (minimum at 4 h: 5.3 ± 0.2 × 10-4 mm²/sec vs. 7.7 ± 0.1 × 10-4 mm²/sec; p < 0.0001).

FIG. 5.

Alterations of diffusivity measures in contusion and pericontusional area. The results are presented in comparison with the contralateral brain tissue of mice subjected to trauma. (A) In the contusion, a significant reduction of mean diffusivity (MD) is visible until 72h after trauma. In the pericontusional area, the MD is significantly higher than contralateral throughout all time points. (B) Correspondingly, free water (FW) signal is low within the contusion but high in the pericontusional area, which is significant compared with contralateral brain tissue for 48h and 7d, respectively. Data are presented as mean ± standard deviation; n = 8 in sham and controlled cortical impact (CCI) group. Two-way repeated measures analysis of variancewith Tukey multiple comparisons test, *p < 0.05, **p < 0.01, ***p < 0.0001 vs. contralateral brain of the trauma group.

By contrast, in the pericontusional area, the MD was significantly higher than in the contralateral brain throughout the whole observation time (maximum 9.8 ± 0.3 × 10⁻⁴ mm²/s vs. 8.1 ± 0.1 × 10⁻⁴ mm²/s at 7 days, p < 0.0001, Fig. 5A). Correspondingly, the FW value indicating the percentage of free water was low within the contusion (minimum 16.1 ± 0.4% vs. 23.0 ± 0.2% contralateral at 4h, p < 0.0001) but high in the pericontusional area (maximum 28.9 ± 1.0 % vs. 24.0 ± 0.2% contralateral at 7 days, p < 0.0001).

This effect was significant within the contusion for 48h and in the pericontusional area for 7 days (p < 0.0001 vs. contralateral brain, respectively, Fig. 5B). These data indicate that the extracellular space is significantly enlarged in the pericontusional brain and significantly reduced in contused tissue, suggesting that mainly VBE is present in the traumatic penumbra while mainly CBE is present in contused brain tissue. At 4h after trauma, the area showing CBE has a size of ca. 36 mm³, which is slightly larger than the actual contusion. The area that shows VBE covers about 25 mm³.

Quantification of vasogenic edema formation by in vivo 2-photon imaging

The relationship between FW and BBB permeability was not investigated before. Therefore, to in vivo verify the results obtained by FWI, we investigated the permeability of the BBB within the traumatic penumbra by in vivo 2-photon microscopy 48h after TBI, the time point where the maximal midline shift was observed (Fig. 3C).

Consistent with FWI, no increased permeability of the BBB was observed in sham operated mice, while a significant extravasation of the plasma dye was detected in ROI 1 and 2 after CCI (Fig. 6A). Here, the fluorescence intensity of the extravasated plasma marker reached 226 ± 60% (p < 0.01 vs. sham) and 187 ± 72% (p < 0.01 vs. sham) of baseline at 120 min after tracer injection, respectively (Fig. 6B, 6C).

FIG. 6.

Vasogenic edema formation assessed by 2-photon microscopy. (A) Representative images of the brain parenchyma in region of interest (ROI) 1 over time. Vessels were labeled with tetramethylrhodamine-dextran (TMRM, 40 kDa). Scale bar = 100 μm. There is no extravasation detectable in sham-operated animals, while extravasation is clearly visible following trauma. (B–D) Quantification of TMRM extravasation in different ROI (1–3). Fluorescent signal intensity was corrected to the results obtained in the sham group at the respective time points to exclude bleaching artifacts. According to the proximity to the contusion, fluorescence intensity increases significantly in A1 and A2, while this is less pronounced in A3. Data are presented as mean ± standard deviation; n = 5 in sham and CCI group. Mann-Whitney Rank Sum test. *p < 0.05, **p < 0.01 vs. sham.

In ROI 3, the area most distant from the contusion, only a modest increase in tissue fluorescence was observed (Fig. 6D), suggesting that vasogenic brain edema was most pronounced in the vicinity of the contused brain.

Quantification of white matter injury after TBI

Although brain edema resolved within one week after trauma as indicated by the normalization of midline shift and hemispheric swelling, MD values continuously increased while the volume of the extracellular space, which is represented by the FW fraction, remained almost unchanged (p < 0.05 vs. 5 days, Fig. 5B, 5C). Therefore, we hypothesized that this increase in water diffusivity is caused by degeneration of white matter tracts, which cause a secondary widening of the extracellular space. To test this hypothesis, we investigated the integrity of the white matter by LFB myelin staining as described previously.^47,48

While white matter tracts showed no pathological changes seven days after sham surgery, TBI reduced the intensity of LFB staining in the traumatized hemisphere by almost 50% (severe) and by 23% in the contralateral hemisphere (p < 0.0001 vs. sham; Fig. 7 A–C). The reduction in LFB staining in the traumatized brain was associated with cavitations within the myelin of the white matter (Fig. 7B), suggesting that the contusion results in an injury of the white matter. These findings indicate that in the absence of brain edema, FWI may be a suitable tool for the quantification of severe white matter injury.

FIG. 7.

Post-traumatic demyelination occurs 7 days after controlled cortical impact (CCI). Representative images of Nissl sections (A) and of Luxol fast blue staining (LFB) (B) at 7 days after trauma or sham operation. (C) High magnification images of LFB staining in the respective region of interest (ROI). Scale bar = 100 μm. White matter cavitation (black arrow) is visible in the traumatized brain at 7 days post-CCI. (D) Quantification of integrated intensity of LFB staining in different areas (ROI 1–ROI 3). After trauma, integrated intensity decreases by approximately 23%, 45%, and 45% in ROI 1, 2, and 3, respectively. Demyelination was more severe in the ipsilateral hemisphere than in the contralateral hemisphere. Data are presented as mean ± standard deviation; n = 8 in sham and CCI group. Two-way repeated measures analysis of variance with Tukey multiple comparisons test, ***p < 0.0001 vs. sham. Au, arbitrary unit.

Discussion

Lesion volume, hemispheric swelling, and midline shift after experimental TBI can be detected by MRI and show similar patterns as previously obtained by histological analysis.^5,38,49 More importantly, our results using FWI demonstrate that both CBE and VBE occur after brain trauma: CBE was observed mainly within contused brain tissue, while VBE dominated in the pericontusional brain. Using FWI, we could further demonstrate widespread white matter injury after TBI.

Because the formation of VBE and white matter injury were validated by two independent methods—i.e., in vivo 2-photon microscopy and LFB staining—our results suggest that FWI represents a reliable and valuable tool for the longitudinal investigation of acute and chronic changes after TBI. Because FWI can also be used on clinical MR scanners, the findings of the current study may have a significant translational aspect.

Several forms of brain edema are described in the literature (e.g., vasogenic, cytotoxic, ionic (osmotic), interstitial (hydrocephalic), and hydrostatic^9,14,17); however, morphologically all forms of brain edema result in an increase or decrease of the volume of the extracellular space (ECS). The ECS volume may decrease by swelling of astrocytes or neurons or increase by influx of water from the intravascular space—i.e., by cytotoxic/ionic edema or VBE.^{4,14,15,50–52} Because changes of the volume of the ECS result in changes of the diffusion of water within the brain parenchyma, already in the 1990s researchers used the apparent diffusion coefficient (ADC) assessed by diffusion weighted MRI to quantify brain edema formation after TBI.^18–20,53

These studies suggested that CBE may be the dominant form of edema after brain trauma.^18–20,53 The MRI techniques available at that time, however, did not allow differentiating between intra- and extracellular water and were thus prone to misinterpretations of the ADC values when CBE and VBE occurred simultaneously (“pseudo-normalization”). Therefore, the ultimate form of brain edema after TBI remained a source of constant debate.

In addition, recent investigations performing in vivo BBB permeability measurements by 2-photon microscopy demonstrated the occurrence of VBE after experimental TBI up to five days after injury.^21,22 To overcome this knowledge gap, we identified FWI to represent a method potentially able to differentiate between vasogenic and cytotoxic edema.^29,32

As expected, no pathological changes were observed in naïve or sham operated animals. After TBI, lesion volume increased during the first 24–48h, which is in accordance with previously published data.⁵ Of note, our first assessment of contusion volume by T2 imaging was at 4h after trauma (not immediately), so the increase in contusion volume was less pronounced. Thereafter, the lesion core consolidated and the lesion shrank by about 60%.

This shrinkage is significant; however, the trend is well in line with previously published data obtained by histological analysis. The lesion expands during the first 24–28h after injury, followed first by a decrease until 7 days after injury and second by another slow increase in lesion volume over 1 year.^{5,38,40,49,54}

Then we analyzed hemispheric swelling and midline shift by T2-weighted imaging. Hemispheric swelling followed the time course of contusion expansion and reached a maximum of 114% of the contralateral hemisphere 24–48h after injury. Thereafter, hemispheric swelling decreased, but the injured hemisphere remained swollen by 5% suggesting that brain edema persisted longer than seven days after TBI, a finding which also can be observed in some patients after TBI.⁵⁵

Hemispheric swelling resulted in a midline shift of about 0.35 mm. After most brain edema regressed, the midline shift fully recovered, a finding fully in line with our current pathophysiological understanding of intracranial volume shifts.

When looking at the MD of water in uninjured brain tissue—i.e., either in naïve or sham operated mice or in the contralateral hemisphere of injured mice—we found remarkably stable intra- and interindividual values that did not change over one week. The same was true for the FW fraction, a parameter essentially reflecting the volume of the extracellular space. For this parameter we recorded values of about 22%, which are very well in line with the known volume of the extracellular space in the mammalian brain of about 20%.⁵⁶

During MRI, the core body temperature of the animals was around 38°C, while FWI is modeled for the diffusivity of water at 37°C.^29.57 Such a temperature change within the physiological range does not cause damage to the restrictive structures present in the brain tissue, and the directionality of the diffusion of water molecules is not significantly affected. Small temperature related variations can be adjusted for by calculation.

Interestingly, we found an increase in MD seven days after TBI—i.e., increased water diffusivity within the traumatized hemisphere, without an increase of FW. We hypothesized that the enlargement of the extracellular space is from white matter injury, a process well known to occur after TBI.^47,48

Using LFB to stain white matter, we identified signs of widespread white matter injury. This phenomenon was strongest in pericontusional brain tissue. We interpret the increase in MD without increase of FW as a hint for strong demyelinization. Thus, next to conventional methods such as DTI, FWI may also serve as an additional tool for the non-invasive and longitudinal investigation of severe white matter injury after TBI or other disorders where severe white matter injury plays a role.

Our data show that the FW fraction—i.e., the extracellular space—within contused brain tissue is already reduced at 4h after TBI, suggesting the formation of CBE. Pial vessels and penetrating arterioles perfusing the cerebral cortex are directly injured by a local high velocity impact as occurs during CCI. Thus, cerebral blood flow to the underlying tissue is almost immediately reduced to ischemic levels and causes cytotoxic cell swelling.⁵⁸ Later, this tissue suffers from massive cell death, is removed, and replaced by a fluid filled cyst.⁵⁹

The data obtained with FWI reflect these well-known changes quite well and, thus, this imaging technique seems to be well suited to differentiate between severely damaged tissue (reduced FW fraction) and normal tissue (normal FW fraction).

The situation changes to the opposite when looking at pericontusional brain tissue. Here, the MD and FW are increased already 4h after TBI, indicating an expansion of the extracellular space as occurs during VBE. Pericontusional tissue is still alive after TBI and may only later be damaged by glutamate toxicity, membrane damage, programmed cells death, or reactive oxygen species.⁶⁰ Thus, opening of the BBB may well be the predominant pathophysiology within pericontusional tissue in the first hours or even days after TBI.

To validate the findings obtained by FWI, we investigated the permeability of the BBB by in vivo 2-photon microscopy and were indeed able to demonstrate BBB leakage in this area, as also shown previously.^21,22 While 2PM provides information with subcellular resolution within a small amount of tissue (1 ROI has 425*425*300μm), MRI provides information on a larger scale for the whole brain.

The pericontusional area in MRI was a “shell” around the contusion with a diameter of 0.92–1.59 mm, and 2-photon data were obtained within this pericontusional area. Hence, our results clearly suggest that the signals obtained by FWI correctly reflect the pathophysiological processes happening within the rim of injured tissue.

Overall, at 4h after trauma, about 59% of edematous brain tissue shows CBE and 41% shows VBE. The CBE occurs mainly where tissue is terminally damaged. At 4h post-trauma, however, the volume affected by CBE is larger than the contusion volume assessed by T2 imaging (ca. 36 mm³ vs. ca. 30 mm³), indicating that CBE is not limited to the primary contusion but also occurs in the ischemic traumatic penumbra.^5,58

The VBE, by contrast, occurs mainly in the tissue at risk, and it develops over a prolonged period–i.e., the therapeutic window is long enough (hours to days) for interventions in a clinical setting. While CBE may be important after brain trauma because it adds to total brain swelling, VBE seems to be an even more relevant and important target for future therapeutic studies.

The FW and MD often change similarly in our results. We also observed differences, however, indicating a benefit in using the two measures in concert. Applying both methods thus provides relevant additional information—for example, in those regions where MD is increased but FW is not. Importantly, this imaging modality was designed for clinical use and was already used in clinical studies investigating alterations such as edema progression⁶¹ or white matter injury after brain trauma.^62–64 Hence, it provides a valuable tool to facilitate and adjust therapeutic strategies in the clinical setting.

Conclusion

Using FWI, we were able to characterize the temporal and spatial profile of brain edema formation after TBI and to identify CBE as the main edema form in contused tissue and VBE as the predominant edema form in pericontusional brain tissue. The FWI may represent a valuable tool for the proper diagnosis and therapeutic management of TBI and may pave the way for further investigations into the mechanisms and the treatment of brain edema formation after TBI.

Transparency, Rigor, and Reproducibility Summary

The study design and analytic plan were ethically approved by the local authorities (government of Upper Bavaria, animal license number Vet_02-23-17). The number of animals investigated in each group was calculated based on a significance level of 0.05, a standard deviation of 25–30% of the mean, a biologically relevant difference between groups of 50%, and a power of 0.8.

33 mice were subjected to experimental injury, 3 were excluded for technical reasons. For the MRI study, 20 animals were randomly assigned to groups by drawing lots (8x trauma, 8x sham operation and 4x naïve) and 1 animal was excluded due to technical reasons. For the 2-Photon study, 5 animals per group were assigned to trauma and sham operation, respectively, and 2 mice were excluded due to technical reasons. Complete data was obtained from 30 animals.

Investigators who performed outcome assessments were blinded to treatment. New studies, which repeat parts of this experiment in the context of our internal quality control, are ongoing. Apart from the free-water toolbox, the MRI tools are available under the following links: https://www.mrtrix.org/; https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/; https://itk.org/.

Authors' Contributions

Conception and study design: SH, NP, SMS. Data acquisition: SH, CE, AW, RIS, FBS. MRI setting: FMB, MD, OP. Data analysis: SH, YG. Data interpretation: SH, NP, SMS, OP. Statistical analysis: SH, SMS. Manuscript preparation: SH, SMS. Manuscript edition: NP. All authors read and approved the final manuscript.

Funding Information

This work was supported by the Deutsche Forschungsgemeinschaft through the Munich Cluster of Systems Neurology (Synergy). Nikolaus Plesnila received support from EraNet Neuron/BNBF (CNSAflame (Grant No. #01EW1502A) and TRAINS (Grant No. 01EW1709). Yinghuimin Guo was supported by the China Scholarship Council (CSC No. 202006230091). Ofer Pasternak received funding via the National Institutes of Health (P41EB015902).

Author Disclosure Statement

No competing financial interests exist.

Supplementary Material

Supplementary Figure S1