An uncontrolled rise in intracranial pressure (ICP) is probably the most common cause of death in traumatic brain-injured (TBI) patients. The ICP rise is most of the time due to cerebral oedema. Different types of oedema coexist in TBI patients: vasogenic oedema and cytotoxic oedema. Vasogenic oedema occurs with the extravasation of fluid into the extracellular space following blood-brain barrier (BBB) disruption. Cytotoxic oedema results from a shift of water from the extracellular compartment into the intracellular compartment due in part to alterations in normal ionic gradients. Several methods have been used to study oedema formation following TBI, but its underlying mechanisms are still not well understood.
Diffusion-weighted imaging is a nuclear magnetic resonance imaging (MRI) technique that provides a useful and non-invasive method for visualizing and quantifying the diffusion of water in the brain associated with oedema. Apparent diffusion coefficients (ADC) can be calculated and used to assess the magnitude of water diffusion in tissues. For example, high ADC values indicate more freely diffusible water and could be considered as a marker of vasogenic oedema. At the opposite, cytotoxic oedema restricts water movement and results in decreased signal intensities in the ADC map. In 1996, Barzo and colleagues  used diffusion-weighted imaging for water content assessment in a rat model of diffuse TBI. The authors studied the temporal evolution of the relative change in ADC after trauma and found an early increase in ADC values during the first 60 min followed by a decrease in ADC values reaching a minimum at 1 week. This result suggests a biphasic oedema formation following TBI, with a rapid and short disruption of the BBB after the first hour post injury, leading to an early formation of vasogenic oedema. However, nuclear MR examination is not advisable at the acute phase in human TBI, especially in unstable head trauma patients. For these reasons, we need to find safer techniques to study oedema formation at the acute phase of TBI.
Computed tomography and estimated specific gravity
Since its development in the 1970s, computed tomography (CT) has remained a radiological examination of choice in the acute assessment of patients with TBI. CT maps the absorption of X-ray by tissues. A crucial point is that the radiological attenuation is linearly correlated with the physical density in the range of human tissue densities [2,3]. For example, blood clot has relatively little water and absorbs more X-ray than the normal brain; it is displayed as a hyperdense area. On the contrary, ischaemia and oedema are displayed in dark areas, because there is an increase in water content. This property gives the unique opportunity to calculate the volume, the weight and the specific gravity (SG) of any tissue by CT. We recently developed a software package (BrainView) developed for Windows workstations, providing semi-automatic tools for brain analysis and quantification from DICOM images obtained from cerebral CT. For each examination, BrainView inputs series of continuous axial scans of the brain. It then automatically excluded extracranial compartments on each section by means of a mathematical morphology-based algorithm. Interactive slice-by-slice segmentation allowed to select different anatomical territories indexed throughout the whole sequence. The software is the upgrade of a software named Lungview, previously developed by the same institution (Institut National des Télécommunications, Evry, France). Lungview was previously validated  and extensively used for lung and heart weight, volume and density analysis by our group [5-7]. This technology allowed one to assess the weight, volume and estimated SG (eSG) of different anatomical parts of the brain hand tracked such as the two hemispheres, the cerebellum, the brainstem and the intraventricular and subarachnoid cerebrospinal fluid (CSF), contused and non-contused hemispheric areas.
For each compartment of a known number of voxels, the volume, the weight and the SG were computed using the following equations:
- Volume of the voxel = surface × section thickness.
- Weight of the voxel = (1 + CT/1000) × volume of the voxel where CT is the CT number of the voxel.
- Volume of the compartment = number of voxels × volume of the voxel.
- Weight of the compartment = summation of the weight of each individual voxel included in the compartment.
- eSG of the compartment = weight of the compartment/volume of the compartment. The eSG is expressed as a physical density in g mL−1.
We validated this technology ex vivo. We first measured the SG of different solutes by determining the weight of 1 L of these solutes. The eSG of the same solutes has also been computed using of BrainView. The two values were linearly correlated, especially in the range of densities in human brain tissue .
Does the brain become heavier or lighter after trauma?
Based on this methodology, we calculated the weight, the volume and the eSG in 15 TBI patients and 15 controls. For a similar age and overall intracranial volume, TBI patients had a brain weight 82 g heavier and hemispheres 91 g heavier than controls. The volume of intraventricular and subarachnoid CSF was reduced in TBI patients. Our results raised new questions: Why does the brain become heavier after trauma? Is it the consequence of water transfer from vessel to the extracellular compartment? Is it due to a leakage of plasma, cells, proteins following BBB disruption after trauma? Is it due to blood engorgement?
One way of finding an answer and to better precise the increased brain weight observed in TBI patients could be to study the eSG in this population. Indeed, in theory, a complete disruption of the BBB with a leakage of water, electrolytes and proteins would increases the brain eSG, since the added volume (exsudat, 1.080 g mL−1) has a density greater than the brain (1.033 g mL−1). On the contrary, a partial disruption of the BBB with a leakage of water and electrolytes would decrease the density, since the added volume (transudat, 1.015 g mL−1) has lower density than the brain (1.033 g mL−1).
Estimated SG as a marker of severity in TBI?
In a first series of measurements in 15 TBI patients 3 ± 2 days after head trauma, we found that the eSG of hemispheres, brainstem and cerebellum was significantly higher in TBI patients as compared with controls (all P < 0.0001)  (Table 1). The increase in eSG was statistically similar in these three anatomical compartments, and in the white and grey matter. Furthermore, there was no correlation between the hemispheric eSG and age, natraemia at CT time, type of accident, initial Glasgow Coma Scale (GCS), presence of a traumatic subarachnoid haemorrhage and presence of intraparenchymal blood.
One could hypothesize that a certain amount of water might be driven out from the interstitial space to the vessels or to the CSF because of cell swelling or increased ICP. Reducing the amount of free water would automatically increase eSG. However, because the overall weight and volume of hemisphere increased, this hypothesis can be ruled out. Natremia of the patients was in the normal range at the time of CT, eliminating a simple difference in osmolarity as a cause of SG increase. One might argue that the increased eSG could be due exclusively to hyperaemia caused by vascular dilation. However, there are some strong experimental  and human data against this hypothesis. Recently, Marmarou and colleagues  demonstrated using MRI that brain oedema is the major fluid component contributing to traumatic brain swelling following TBI in man. These authors observed a reduction in cerebral blood volume (CBV) in proportion to cerebral blood flow following severe brain injury. A similar result was observed using emission CT of 99mTc-labelled red cells . In this study, CBV was reduced soon after injury, preferentially at the expense of the grey-matter compartment. As shown on the abacus presented in Figure 1, an increase in CBV of 45 mL would be theoretically necessary to increase hemispheric SG from the mean value of controls to the mean value of TBI patients, since blood has an SG of 1.060 g mL−1 . Considering that normal CBV is about 5% of the overall intracranial volume, this would mean a 65% increase in CBV. Together, the mean change of hemispheres volume that we observed was 85 mL, a value much higher than what could be explained by the change in CBV alone. The hypothesis of an increase in eSG due to a traumatic subarachnoid haemorrhage also does not fit our data. The volume added in hemispheres of our patients should have a density of 1.045 g mL−1 to explain our results (considering the mean values). This value is not compatible with a leakage of plasma alone, it must also involve cells.
To confirm our first observation and better understand the significance of the elevated eSG observed in TBI patients, we decided to perform a new study including a much greater number of patients and to restrict the analysis to the initial CT only performed in the first 5 h after trauma.
In all, 120 TBI patients were included in this new study and divided into two groups according to the initial eSG of the non-contused hemispheric areas. In the normal SG group, patients have an eSG less than 1.96 SD above controls. On the contrary, in the increased SG group, patients have an eSG higher than 1.96 SD.
The results of this second study confirmed our first observation on brain eSG in human TBI using the same CT methodology but in a larger group of patients and at a different time-window after trauma. In addition to this initial observation, we classified patients into two groups according to their initial SG. Patients with the highest SG had a lower GCS and more often a mydriasis on the scene of the accident, received more frequently osmotherapy at the initial phase, had more frequently an extra ventricular drainage implanted for ICP monitoring and CSF drainage, received more frequently barbiturates as a second-line therapy and had more frequently a CT classified in the third category of the Marshall Score. The initial GCS and simplified acute physiological score (SAPS 2), the velocity, the occurrence of mydriasis at the scene and the use of osmotherapy were predictors of outcome at ICU discharge and at 1 yr. Only the SAPS 2 and the use of osmotherapy remained significant predictors in multivariate analysis. eSG of the overall intracranial content or of the non-contused areas were also predictors of outcome, but at ICU discharge and in the monovariate analysis only. This indicates that long-term outcome might be independent of the intensity of the initial brain oedema, providing that efficient therapies to treat increased ICP are applied rapidly.
The observation of an increased eSG in patients with TBI is in strong opposition with the data derived from the experimental literature. Most of the studies performed in models of head trauma report a decrease in eSG with a rise in the cerebral water content [13,14]. However, and as in our studies, Nath and colleagues  also observed an increased eSG in eight patients with severe TBI. eSG was determined on small pieces of subcortical using a graduated SG column at 1.0325 ± 0.0012 g mL−1, a value much higher than normal . Similar increase in eSG in head trauma patients has also been observed by Bullock and colleagues  in the same proportion. Theoretically, a leak of plasma decreases the overall hemispheric SG, since the SG of plasma (between 1.0245 and 1.0285 g mL−1) is lower than the one of the brain. On the contrary, a leakage of plasma and cells should increase eSG. This phenomenon is illustrated in Figure 1. According to this abacus, the volume added in hemispheres of our patients should have a density of 1.045 g mL−1 to explain our results (considering the mean values). This value is not compatible with a leak of plasma alone but must also involve cells. The clinical usefulness of the automatic determination of SG in human TBI will have to be addressed in a large prospective study.
The observation of an increased eSG in patients with TBI is in strong opposition with the data derived from the experimental literature. This increase is diffuse and occurs in normally appearing areas on CT Dynamic. This diffuse and early increase in eSG could result from an early initial insult of the BBB at the time of trauma. Patients with an increased eSG received more frequently osmotherapy at the initial phase, had more frequently an external ventricular drainage implanted for ICP monitoring and CSF drainage, received more frequently barbiturates as a second-line therapy and had more frequently a CT classified in the third category of the Marshall Score. The clinical usefulness of the automatic determination of eSG in human TBI will have to be addressed in a large prospective study.
1. Barzo P, Marmarou A, Fatouros P, Corwin F, Dunbar J. Magnetic resonance imaging-monitored acute blood-brain barrier
changes in experimental traumatic brain injury
. J Neurosurg
2. Phelps ME, Gado MH, Hoffman EJ. Correlation of effective atomic number and electron density with attenuation coefficients measured with polychromatic X-rays. Radiology
3. Mull RT. Mass estimates by computed tomography: physical density from CT numbers. Am J Radiol
4. Puybasset L, Cluzel P, Chao N, Slutsky AS, Coriat P, Rouby JJ. A computed tomography scan assessment of regional lung volume in acute lung injury. The CT Scan ARDS Study Group. Am J Respir Crit Care Med
5. Puybasset L, Gusman P, Muller JC, Cluzel P, Coriat P, Rouby JJ. Regional distribution of gas and tissue in acute respiratory distress syndrome. III. Consequences for the effects of positive end-expiratory pressure. CT Scan ARDS Study Group. Adult Respiratory Distress Syndrome. Intensive Care Med
6. Rouby JJ, Puybasset L, Cluzel P, Richecoeur J, Lu Q, Grenier P. Regional distribution of gas and tissue in acute respiratory distress syndrome. II. Physiological correlations and definition of an ARDS Severity Score. CT Scan ARDS Study Group. Intensive Care Med
7. Rouby JJ, Puybasset L, Nieszkowska A, Lu Q. Acute respiratory distress syndrome: lessons from computed tomography of the whole lung. Crit Care Med
8. Lescot T, Bonnet MP, Zouaoui A et al.
. A quantitative computed tomography assessment of brain weight, volume, and specific gravity in severe head trauma
. Intensive Care Med
9. Kita H, Marmarou A. The cause of acute brain swelling after the closed head injury in rats. Acta Neurochir Suppl (Wien)
10. Marmarou A, Bandoh K, Yoshihara M, Tsuji O. Measurement of vascular reactivity in head injured patients. Acta Neurochir Suppl (Wien)
11. Kuhl DE, Alavi A, Hoffman EJ et al.
. Local cerebral blood volume in head-injured patients. Determination by emission computed tomography of 99mTc-labeled red cells. J Neurosurg
12. Blacque-Belair A, Mathieu de Fossey B, Fourestier M. Dictionnaire des Constantes Biologiques et Physiques
. Paris: Maloine, 1965.
13. Gardenfors A, Nilsson F, Skagerberg G, Ungerstedt U, Nordstrom CH. Cerebral physiological and biochemical changes during vasogenic brain oedema
induced by intrathecal injection of bacterial lipopolysaccharides in piglets. Acta Neurochir (Wien)
: 601-608; discussion 608-609.
14. Talmor D, Shapira Y, Artru AA et al.
. 0.45% saline and 5% dextrose in water, but not 0.9% saline or 5% dextrose in 0.9% saline, worsen brain edema two hours after closed head trauma
in rats. Anesth Analg
15. Nath F, Galbraith S. The effect of mannitol on cerebral white matter water content. J Neurosurg
16. Takagi H, Shapiro K, Marmarou A, Wisoff H. Microgravimetric analysis of human brain tissue: correlation with computerized tomography scanning. J Neurosurg
17. Bullock R, Smith R, Favier J, du Trevou M, Blake G. Brain specific gravity and CT scan density measurements after human head injury. J Neurosurg