The mortality statistics of severe traumatic brain injury (TBI) have changed very little in the last 20 years. The principal mechanism leading to death is uncontrolled intracranial hypertension and the ischemic brain damage it produces.1 The cause of such intracranial hypertension is the increase in brain volume triggered by different mechanisms, including vasogenic or extracellular brain edema secondary to rupture of the blood–brain barrier (BBB) and cytotoxic or intracellular edema, associated to ischemia. Intracellular edema is considered to be the main contributor to brain swelling after closed trauma.2–4

Diffusion-weighted magnetic resonance imaging (DWMR) produces a signal with an intensity proportional to the diffusion of water molecules in the tissues. The signal depends on the diffusion capacity of the molecules, on the interactions among them, and on the obstacles facing molecular movement. The intensity of the signal can be quantified by the apparent diffusion coefficient (ADC), which is calculated by means of the following formula: ADC=(1/b)log(S1/S0), where “b” is a factor that depends on the strength and duration of the pulses applied to increase diffusion, “S1” is the intensity of the diffusion-weighted signal, and “S0” is the intensity of the non-diffusion-weighted signal. If this formula is applied to each voxel (three-dimensional graphic element), the end result is a map or image in which the intensity of the signal is proportional to the ADC of each region or tissue examined.5–8

Conventional magnetic resonance imaging (MRI) is unable to distinguish between types of edema based on the intensity of the signal; both types of edema, cytotoxic and vasogenic, produce a hyperintense signal in T2-weighted sequences. In contrast, DWMR differentiates cytotoxic edema, which is characterized by an increase in intracellular fluid uptake, resulting in low ADC values. The ADC is reduced as a consequence of restricted water molecule diffusion produced by the membranes and organelles within the cell, and by the increase in viscosity.

In contrast, in vasogenic edema, the fluid displaces from the vessels to the extracellular compartment, which therefore increases in size and allows the molecules to move more freely. This produces different effects in the magnetic resonance image: either high signal intensity in the ADC and DWMR maps or, less frequently, high signal intensity in the ADC map and low signal intensity in the DWMR image (facilitated diffusion). These changes in signal intensity are generally located around a focal parenchymal lesion, for example, a hematoma, since the blood clot exerts an osmotic effect upon the surrounding brain tissues.4,9

To date, the main application of DWMR has been the detection of acute cerebral infarction. In this context, it is able to detect stroke earlier than any other technique, thanks to its high sensitivity, and DWMR is moreover able to predict the infarct size and assess the patient prognosis.10–13

DWMR has been used in patients with TBI to evaluate diffuse axonal injury (DAI), in view of its greater sensitivity compared with other conventional techniques (T2 and FLAIR sequencing) in the acute phase after trauma.14,15 In addition, the volume of the lesion detected with this technique is correlated to clinical outcome.

Since DWMR is able to distinguish between extracellular and intracellular edema, it could be used to evaluate the contribution of both types of edema to the damage observed in patients with severe TBI.16 Furthermore, the early detection of secondary ischemic lesions in severe TBI would be of great importance, given their impact upon morbidity-mortality.

The present study analyzes the ADC values in a series of patients with focal trauma and their correlation to initial clinical severity and ultimate patient prognosis or outcome.

We recorded 71 patients with severe TBI admitted to the ICU between November 2001 and October 2004. DWMR was performed in 23 of them.

Tables 1 and 2 show the epidemiological and clinical data.

Regarding the types of injuries, surgically evacuated space-occupying lesions were the most frequent presentation, followed by type II lesions. There were no type III diffuse lesions in our series. Seven lesions were defined as diffuse axonal injury (DAI) on the basis of their magnetic resonance location in the subcortical white matter, corpus callosum or brainstem, together with persistent coma (Glasgow coma score<9).

Regarding outcome as determined by the GOS, 70% of the patients had died or were in a vegetative state or seriously disabled (poor prognosis) at discharge from the ICU, versus 43% after 6 months. These results are reported in Table 3.

Imaging data

Attempts were made to perform MRI as soon as possible after admission, though in only four patients was the exploration made within the first 24h, while 12 patients (14 lesions) were subjected to MRI evaluation within the first 48h. In the rest of the cases the timing of the study was highly varied due to different reasons–fundamentally patient instability and the distance to the MRI unit, which according to the physician in charge questioned the usefulness of the study. In any case, we finally decided to analyze all the studies, even if performed over a broad interval of time. Globally, MRI was performed at an average of 101.39±111.80h (range 10–360) after trauma.

We selected 26 ROIs for analysis in the images of the 23 patients, and documented the ADC in each of them. Table 4 shows the mean values globally and according to the type of lesion. An increase in diffusion in the non-evacuated space-occupying lesions was observed when compared with the rest of the types of injuries and with the normal values.

Table 4.

Apparent diffusion coefficient, global and according to type of injury.

Type of injury		ADCa	Normal mean ADC
Global	No.=26	957.80±194.88b
Space-occupying lesions	No.=16	985.3±197.7	890±40 (780–1090) in the cortical gray matter
Untreated hemispheric hematoma or contusion	No.=12	1029.91±127.39
Surgically evacuated hemispheric hematoma or contusion	No.=4	851.75±322.76
Unilateral diffuse injury	No.=3	876±250.21	890±40 (780–1090) in the cortical gray matter
Diffuse axonal injury or brainstem contusion	No.=7	929.85±182.70	700±30 (620–790) in the white matter730±30 (670–820) in the thalamus

a

ADC: apparent diffusion coefficient.

b

ADC units: mm2/s×10−6. The table reports the mean±standard deviation.

The ADC values showed a progressive increase with the time from trauma–ADC in the first 48h being lower than the values recorded later in time (898.7±205.3/1026.6±164).

The prognosis (outcome) was considered poor if the patient had died after 6 months or was in a vegetative state or seriously disabled, and good in the case of patients with moderate or slight disability, or no disability. As can be seen in Table 5, the patients with a poor prognosis had lower ADC values than those with a good prognosis. Fig. 1 shows the outcomes according to the ADC findings at the three measurement timepoints. It should be noted that both in the global results and on considering only the measurements made in the first 24h and those made in the first 72h, the ADC values were always lower in the patients with a poor prognosis–though the sample size was too small to allow the drawing of firm conclusions.

Table 5.

Mean ADC value in relation to patient outcome.

GOS 6 months	ADC: mean±SD (95%CI)		No. lesions
	Poor outcome	Good outcome
All injuries	901.3±231.9 (753.9–1048.6)	1006.2±148.5 (920.4–1.091.9)	26
Diffuse axonal injury	916.6±191.4 (441.0–1392.2)	939.7±204.9 (613.5–1.265.9)	7
Hemispheric injuries	896.2±254.2 (700.7–1091.6)	1032.8±122.9 (944.8–1.120.7)	19
Total	12	14

95%CI: 95% confidence interval of the mean; SD: standard deviation. Nonsignificant (Mann–Whitney U-test). ADC units: mm2/s×10−6.

Figure 1.

Patient outcome according to the ADC findings, globally and corresponding to the three measurement timepoints.

(0.11MB).

• 1.

  The diffusion of water in the brain tissues, measured by MRI techniques, was greater in contusions than in other types of injuries, which could be an expression of vasogenic edema.

  Findings of other studies: regarding the changes in ADC after traumatism, experimental studies in animals have yielded contradictory results, as can be seen in Table 6, where some authors have reported an increase in diffusion as a result of vasogenic edema21 while others describe a two-phase response22 or even significantly diminished diffusion values.23 Some studies only report changes when secondary damage is added, with diffusion decreasing as a result of cytotoxic edema.24–29

  Table 6.

  Studies in animal models of brain injury.

  Author/year	Model of brain injury	Experimental protocol	Results
  Hanstock (1994)21	Percussion	18 rats: 9 in the trauma group, 3 in the unilateral carotid ligation group, and 6 controls.The images were obtained 1, 2 and 4h after ischemia or trauma, and the ADC values were calculated	A significant increase in ADC was observed in the damaged cortical gray matter, 60min after trauma. In rats with ischemia, ADC was significantly decreased at this timepoint
  Alsop (1996)23	Percussion	10 rats: 6 in the trauma group and 4 controlsDWMR after injury	The ADC values decreased in the first 45min
  Ito (1996)24	Impact-acceleration	36 rats in 4 groups: control, trauma alone, hypoxia and hypotension, and trauma with hypoxia and hypotension (THH)DWMR at intervals of 1h after injury	ADC in the THH group decreased quickly to a minimum 3h after injury. In trauma lone, ADC did not decrease significantly
  Barzo 199722	Impact-acceleration	36 rats: 30 in the trauma group and 6 controlsADC and brain water content were measured immediately after injury and on days 1, 3, 7, 14, using spin-echo or SVSa diffusion-weighted imaging	ADC increased in the first hour and then decreased, reaching a minimum on day 7
  Barzo (1997)28	Impact-acceleration	50 rats (42 trauma and 8 controls). ADC and brain water content were measured immediately after injury and on days 1, 3, 7, 14, using spin-echo or SVSa diffusion-weighted imaging	ADC increased in the first hour and then decreased, reaching a minimum on day 7
  Beaumont (2000)26	Impact-acceleration	44 rats: Group 1; trauma with hypoxia and hypotension. Group 2; trauma aloneEffect of the addition of dopamineMeasurement of ADC at baseline and after 4h	Dopamine worsened swelling
  Portella (2000)25	Impact-acceleration+secondary injury	Groups of rats: controls, moderate trauma with/without hypoxia and hypotension, severe trauma with/without hypoxia and hypotension, hypoxia, and hypotension aloneSpin-echo or SVSa diffusion-weighted imaging	Increased cortical water content was observed 4h after injury, inversely related to the change in ADC Secondary injuries act synergically with trauma, increasing intracellular water
  Van Putten (2005)29	Percussion+secondary injury	4 groups: controls, hypoxia alone, trauma alone, and hypoxia and traumaADC mapping 2, 4, 24h and 1 week after injury	Early decrease in ADC, followed by an increase after one week. No changes related to hypoxia were observed. Good correlation to histopathological changes

  a

  SVS: single voxel spectroscopy.

  The apparently conflicting results of these studies could be attributed to the diversity of lesion models used, the different MRI techniques employed for measurement, and the lack of distinction between cortical lesions and diffuse axonal injury, which could involve different mechanisms and evolutive patterns. In animal studies, the ADC values are generally measured in the first minutes or hours after injury, while in humans ADC is measured from hours to days after trauma in many cases. This could explain the different ADC variation profiles observed.

  Human studies have focused on three aspects: the identification of diffuse axonal injury and its diffusion characteristics; analysis of the type of edema in the brain lesions; and evaluation of the usefulness of DWMR in predicting the clinical outcome. Many authors agree that DWMR is more sensitive than other conventional MRI techniques in detecting DAI,30–32 but the observed diffusion changes vary. In a study of 116 patients with DAI, 64% of the lesions had diminished ADC values while 34% presented increased values.31 Liu reported decreased diffusion values in 9 patients with DAI. In these patients, the ADC values decreased significantly in typical DAI observed in conventional MRI images, from as early as one day after injury, and this decrease persisted for up to 18 days.33 On the other hand, the significance of these changes is also controversial. In this context, some authors consider that lesions with low ADC values are indicative of severe brain damage and could predict the long-term outcome.30,32 Schaefer et al.32 studied 26 patients with DAI using DWMR. They observed a strong correlation between the volume of the signal intensity alteration in the diffusion-weighted images and initial clinical severity, but this correlation did not improve on only considering lesions with decreased ADC values. Hou et al.34 in turn analyzed the changes in diffusion in 37 adults with brain injuries. Patients with more serious injuries were found to have ADC values significantly different from those of the controls, and patients with an unfavorable prognosis presented significantly higher ADC values than individuals with a favorable prognosis or the controls.

  In our series we identified 7 cases of DAI, with mean ADC values above normal. After 6 months, three of these patients had a poor prognosis.

  Regarding the type of edema, some authors consider that the distinction between intra- and extracellular edema is of key prognostic and therapeutic importance.35–37 In a series of patients with traumatisms in which the fluid content and ADC were measured using MRI, the authors observed an increase in tissue fluid associated with a decrease in ADC–thus suggesting predominant intracellular edema,35 though in this study the type of lesion and the time elapsed between trauma and MRI were not specified. In a later study, the same group36 used MRI techniques to identify the type of edema that developed in 44 patients with serious brain injuries. The patients with brain swelling showed an increased brain water content and low ADC values, compared with the healthy volunteers. The authors concluded that traumatic brain swelling appears to be predominantly cellular, as suggested by the low ADC values and the high tissue water content.

  Our patients predominantly presented mass-type lesions, and the latter (12 non-evacuated, mixed-density contusions) showed an increase in diffusion that could be an expression of vasogenic edematization, even when only considering the images obtained in the first 48h after injury. Such lesions would correspond to type I injuries according to the classification proposed by Gasparetto,4 referred to traumatic brain injury patterns in DWMR/ADC imaging. According to this classification, type I injuries consist of hyperintense images in diffusion-weighted imaging with an increase in ADC, representing vasogenic edema. Type II injuries in turn comprise hyperintense lesions in diffusion but with a decrease in ADC, indicating cytotoxic edema, while type III injuries correspond to central hemorrhagic lesions surrounded by a hyperintense area in the diffusion image and an increase in ADC. It is important to note that in diffusion imaging of hemorrhagic lesions it is not clear whether low ADC values correspond to cytotoxic edema or to artifacts produced by extravasated blood.

• 2.

  Our patients with a poor outcome had lower ADC values than those with a good outcome. Many studies have described different ADC values in patients with a poor prognosis compared with controls or patients with a good prognosis, but there is little consensus regarding the pattern of change, as reflected by the abovementioned studies.31,32,34,36–39

• 3.

  Study limitations.

  Of all the severe brain injury patients admitted to the ICU, we were only able to use DWMR in 23 cases, due to problems related to critical patient transfer, which make it difficult to perform early and repeated MRI scans. As a result, our sample size was small and scanning covered a broad time interval.

  In addition, we had no cases of diffuse swelling for comparison with other types of injuries; as a result, our study fundamentally centers on focal injuries.

  Another recognized limitation is the use of reference values cited in the literature, instead of values referred to contemporaneous controls.

The mean ADC values in the lesions were greater than the normal values and greater in contusions than in other types of injuries, thus indicating that extracellular edema probably plays a dominant role in these lesions.

The ADC values were decreased in patients with a poor prognosis, suggesting an association between ischemia and patient outcome.

The authors declare that they have no conflicts of interest.