In vivo Longitudinal Diffusion Tensor Imaging of Spinal Cord Injury in Rats

A. A. Deo, R. J. Grill, K. M. Hasan, P. A. Narayana Biomedical Engineering, University of Houston, Houston, Texas, United States, Department of Neurosurgery, University of Texas, Houston, Texas, United States, Diagnostic and Interventional Imaging, University of Texas, Houston, Texas, United States Introduction: Damage to white matter fiber tracts is a major consequence of traumatic spinal cord injury (SCI) and results in functional deficit. Diffusion tensor imaging (DTI) is a powerful modality for noninvasively following temporal changes in the fiber tract pathology. With the exception of a single abstract [1], DTI studies of SCI in rodents are limited to ex vivo [2], [3], which are limited to providing only a snap shot picture at a single time point. In addition, in vivo DTI studies of normal spinal cord, have clearly demonstrated that in vivo diffusion measures are quantitatively different from those of excised cords [4]. Majority of the published DTI studies of SCI focused on the anisotropic indices for assessing the tissue integrity. Recent studies strongly suggest that the individual diffusivities (eigenvalues) along the length of the cord (λl,) and perpendicular to it (λt,) provide improved pathologic specificity compared to the diffusion anisotropic indices [5]. Here we report the in-vivo longitudinal changes in the DTI metrics that include mean diffusivity (Dav), fractional anisotropy (FA), and individual eigenvalues of different white matter regions for up to 8 weeks post-injury and demonstrate regionally dependent pathologic changes. Methods: All MR studies were performed on a 7-T Bruker scanner. A total of twelve male Sprague-Dawley rats weighing between 300 and 350 g were used in these studies and divided into two groups of six each. One group received moderately severe injury at the T7 level while the second group underwent sham operation without injury and served as the control. A 11x27 mm rectangular radio frequency coil was implanted over the site of injury and coupled inductively to an external coil of 15x30 mm in size. The BBB scores were determined every week. DTI data was acquired at two, four, six and eight weeks post injury using a multi-shot EPI sequence from 20 contiguous axial slices, each of 2 mm thick with an in-plane resolution of 200 microns and square FOV of 26.2 mm. The DTI metrics, FA, λl, λt, and Dav, were determined from dorsal, ventral, lateral (D-, V-, L-) white matter that are rostral and caudal to the epicenter of the injury. ANOVA with repeated measures was performed for comparing the differences between mean FA, λl, λt, and Dav values for all the three regions at each time point. Results: The temporal changes of the DTI metrics 2 mm rostral and caudal to the site of injury are shown in Table 1. The FA values of injured cord are lower and Dav values are higher compared to the uninjured normal cord at all time points. Since the injured cord showed decreased λl and increased λt, relative to normal at all locations and at all time points, we considered an increase in λl and a decrease in λt as an indication of tissue recovery. Rostral section: V-WM showed an increase in the FA value from day 14 to 28, followed by a decline on day 56. λt, initially showed a decrease, followed by an increase on day 56. L-WM showed an initial increase in FA from day 14 to 56. λt, continued to decline until day 56. Temporal changes were not observed in either Dav or λl in both V-and L-WM. Caudal section: V-WM showed an initial decline in Dav from day 14 to day 28 followed by an increase up to day 56. λl remained constant until day 42, but increased by day 56. Statistically significant temporal changes were not observed in FA and λl. L-WM showed an increase in FA until day 42, followed by a decline by day 56. The other DTI metrics did not show any temporal changes. D-WM did not show any significant differences with time in both rostral and caudal section. The results at 4 mm caudal and rostral sections were very similar to those observed at 2 mm, except that the temporal changes in different region are smaller. At 6 mm, the DTI metrics were not statistically different from the uninjured tissue at all time points investigated in these studies.Statistically significant correlation was only observed between the BBB scores and λl in the L-WM in the rostral section. No correlation was observed between BBB and of the DTI metrics in the caudal section. Discussion and Conclusions: The DTI metrics show tissue disruption in regions that appear normal on the conventional MRI, suggesting the potential of DTI in detecting subtle pathology in SCI. These studies also demonstrate that the disruption of white matter tracts is less pronounced with increased distance from the epicentre. Based on the published studies [5], demyelination increases the eigenvalue representing the transverse water diffusion, λt, with relatively little effect on the longitudinal water diffusion, λl. On the other hand, axonal damage without demyelination appears to result primarily in reduced λl. Taken together, our results suggest recurrence of demyelination after day 28 in the V-WM and continued remyelination in the L-WM. These results also suggest lack of axonal recovery in all the three regions of the white matter in the rostral sections. Our studies demonstrate a very heterogeneous tissue recovery in different regions of the injured cord and showed caudal-rostral asymmetry in the tissue recovery. Finally, the individual diffusivities appear to improve the pathologic specificity over the anisotropic index alone. Rostral: V-WM FA λt D14 0.64 ± 0.045 0.65 ± 0.033 D28 0.70 ± 0.085 0.52± 0.030 D42 0.68 ± 0.059 0.58 ± 0.036 D56 0.65 ± 0.097 0.65 ± 0.014 [14, 28] P=0.04 [14, 28] P=0.006 [28, 56] P=0.01 [28, 56] P=0.005 Rostral: L-WM FA λt D14 0.64 ± 0.053 0.75 ± 0.056 D28 0.65 ± 0.045 0.68 ± 0.046 D42 0.69 ± 0.084 0.58 ± 0.019 D56 0.68 ± 0.047 0.61 ± 0.049 [14, 42] P=0.02 [14, 42] P=0.027 [14, 56] P=0.01 [14, 56] P=0.048 [28, 42] P=0.01 Caudal: V-WM Dav λt D14 1.01 ± 0.09 0.49 ± 0.054 D28 0.95 ± 0.15 0.46 ± 0.039 D42 0.99 ± 0.12 0.44 ± 0.054 D56 1.11 ± 0.19 0.54 ± 0.068 [28, 56] P=0.04 [42, 56] P=0.02 [42, 56] P=0.04 Caudal: L-WM FA D14 0.66 ± 0.121 D28 0.71 ± 0.111 D42 0.72 ± 0.110 D56 0.70 ± 0.080 [14, 28] P=0.03 [14, 42] P=0.01 [14, 56] P=0.01