After traumatic brain injury (TBI), damaged brain tissue may demonstrate different degrees of hemorrhage, ischemia, or infarction. Surgical evacuation is often needed for lesions causing mass effect, but guidelines for the extent of tissue removal are unclear. In this study, we used positron emission tomography (PET) to study the tissue viability in pericontusional regions.
Methods
Eight TBI patients (male; GCS 8–12) with intracerebral contusions were studied. In addition to routine CT scans, MR and PET scans, taken within four days after injury, were co-registered. Grey matter (GM) and white matter (WM) regions were separated using a MR-based segmentation technique. To analyze the pericontusion regions, we used a semi-automatic method to separate the contusional core and adjacent edema from the normal-appearing tissues using the conformal anatomical information derived from CT/MR images and cluster analysis. Metabolic changes occurring in contusions and surrounding tissues were comprehensively characterized using quantitative PET studies measuring regional cerebral blood flow (rCBF), oxygen extraction faction (rOEF) and oxygen metabolism (rCMRO2). The coupling between flow and metabolism was evaluated by examining their correlations (voxel-wise rCBF vs. rOEF and rCBF vs. rCMRO2, respectively). Voxel values of rCMRO2 in sixteen normal subjects were studied to determine a threshold value of living tissue for TBI data analysis.
Results
The pericontusional region clustered into four distinct groups (e. g. Fig. 1A CT image, B: cluster image). One cluster (in black; Fig. 1A) corresponded to the CT hyper-density contusional core and another (in dark gray) matched the surrounding CT hypo-density edematous tissue. PET data showed that the normal-appearing GM brain tissue adjacent to but 1 cm rim away from the hypo-density pericontusional region on CT had concurrent decreased flow/metabolism with a preserved ratio of rCBF to rCMRO2. The pericontusional region, however, had a variable flow/metabolism profile characterized by a nonviable core (i.e. CT hyper-density voxels) surrounded by damaged tissue (i.e. CT hypo-density voxels). The degree of flow/metabolic abnormalities centrifugally decreased with increasing distance from the core (e.g. labeled as group 1, 2 and 3, respectively, in Fig. 1C; the voxels correspond to the dark gray cluster within the circular area of Fig. 1B and were separated into three groups using two rCMRO2 threshold values). Within the pericontusional CT hypo-density regions, 44 ± 19% (mean ± 1 s.d.; range 11–60%; n = 8) of the GM voxels had rCMRO2 values less than the 5%ile of normal WM rCMRO2 values (i.e. 0.46 ml/min/100 g). In contrast, 22 ± 11 % (range 5–40%) of the GM voxels had rCMRO2 values that were similar to (within 80% confidence interval) the corresponding TBI normal-appearing GM values.
Conclusion
Our regional metabolic data suggest that the CT hyper-dense core and more than 10% of the adjacent hypo-dense area are severely damaged and likely nonviable (CMRO2 values below 0.5 ml/min/100 g). Surgical evacuation of severely damaged tissue immediately surrounding contusions is not expected to be harmful to the patient.