Abstract
Spinal cord injury (SCI) is a dramatic condition with serious consequences. Patients suffer from sensorimotor deficits and a plethora of associated complications. Functional recovery varies tremendously between individuals and the exact factors determining the recovery potential are not fully understood. The initial neurological impairment already provides a good estimate of the SCI severity but it does not explain the entire recovery potential as patients with similar impairments still show distinct recovery trajectories. Thus, additional biomarkers for improving prognostication following SCI are needed. A crucial factor for the development of more accurate prognostic markers is a profound understanding of the pathophysiological processes. It has already been shown that a spinal trauma is not restricted to the site of the initial injury but that secondary neurodegenerative changes (i.e., pathophysiological consequences of the injury) affect the entire neuraxis from cortical brain areas above the injury to the lumbar cord below the injury. However, the exact pathomechanisms underlying this extensive neurodegeneration and their dynamics are understudied. A non-invasive and clinically very useful technique to study alterations in neural tissue is magnetic resonance imaging (MRI). It allows to visualize the damage of the spinal trauma and to quantify its extent. With the availability of advanced MR protocols, which include quantitative MRI, functional MRI, and MR spectroscopy, even the tissue's microstructure, functional connectivity, and metabolic environment can be studied. These techniques show great potential for the development of new and very sensitive biomarkers to assess pathological changes following SCI and predict the functional outcome. The overall goal of this thesis was to improve our understanding of SCI and its consequences by optimizing MR acquisition protocols, investigating micro- and macrostructural as well as metabolic changes across the entire neuraxis, and by assessing the predictive value of MRI-derived biomarkers. In the first study, we investigated pathophysiological changes following SCI in the upper cervical cord in terms of changes in macrostructure (i.e., tissue atrophy) and microstructure (i.e., demyelination) using structural and quantitative MRI. In particular, we assessed the spatiotemporal dynamics of retrograde and anterograde neurodegenerative processes and tracked the development of a spatial neurodegenerative gradient over two years after SCI. Tissue atrophy and demyelination indicated both retrograde and anterograde neurodegeneration in the cervical cord. However, while anterograde neurodegenerative changes were detectable uniformly across the cervical cord very early after injury, retrograde neurodegenerative alterations gradually developed over two years and showed a spatial gradient across the cord with more pronounced changes closer to the lesion. This demonstrates fundamental differences in the spatiotemporal dynamics of both processes. Crucially, microstructural markers of myelination were more sensitive to early changes compared to macrostructural markers and allowed prediction of neurological recovery. The overall objective of studies II and III was to reduce the acquisition time of imaging protocols for patient benefit by utilizing synthetic MRI. This technique allows to reconstruct various contrast-weighted images from quantitative maps of tissue properties without the need for additional structural scans. Specifically, in study II, we investigated the feasibility and precision of synthetic T1-weighted MRI for measuring spinal cord atrophy following SCI. We demonstrated an excellent test-retest repeatability for synthetic MRI, which was within the range of acquired MRI and showed the validity of synthetic MRI for tracking cervical cord neurodegeneration after SCI. Thus, synthetic MRI can be considered as a valid alternative to acquired MRI for reducing the acquisition time of imaging protocols. However, synthetic MRI showed a systemic bias in the measurements of the cross-sectional spinal cord area compared to acquired MRI. Therefore, we conducted study III, which aimed at optimizing the reconstruction parameters of synthetic MRI with respect to accuracy and reducing the systemic bias compared to acquired MRI. We showed that by optimizing the reconstruction parameters, the accuracy of synthetic MRI can be improved considerably allowing for precise and accurate measurements. Moreover, synthetic MRI showed smaller variance in repeated measurements over time resulting in a considerable reduction in the required number of participants for prospective studies. Thereby, synthetic MRI can help to further optimize imaging protocols of clinical studies. In the fourth study, the relevance of the level of injury for the extent of the focal intramedullary lesion and for prognostication was assessed. We compared the focal lesion extent between cervical and thoracolumbar SCI patients and demonstrated that both patient groups differed particularly in the craniocaudal extent of the lesion while anteroposterior parameters were less sensitive to differences between cervical and thoracolumbar SCI. Crucially, anteroposterior parameters, in particular the amount of preserved tissue around the lesion, showed the best prediction of neurological recovery, which highlights the utility of these parameters as neuroimaging biomarkers for prognostication after SCI. The fifth study aimed at investigating metabolic changes in neural tissue following SCI to elucidate the pathomechanisms of remote neurodegeneration across the neuraxis. We applied proton MR spectroscopy in both the motor cortex and the lumbar enlargement and found reductions in metabolic markers of neuronal integrity and density. This suggests neuronal atrophy occurring in both the motor cortex and the lumbar cord as a consequence of retrograde neurodegeneration and trans-synaptic neurodegeneration, respectively. These results demonstrate in-vivo the remote effects of SCI for the metabolic environment and the pathomechanisms leading to macrostructural tissue atrophy in the brain and the lumbar cord. In conclusion, a multimodal imaging approach was applied in this thesis to disentangle the pathophysiological changes following SCI that affect the entire neuraxis with the ultimate goal of improving patients' quality of life. First, we contributed towards the optimization of imaging protocols in clinical studies by demonstrating the feasibility and validity of synthetic MRI for tracking neurodegeneration after SCI. Next, we extended our understanding of the remote consequences of SCI using structural and quantitative MRI as well as MR spectroscopy by showing macrostructural atrophy and demyelination in the cervical cord as well as neuronal atrophy in the brain and the lumbar cord. Finally, we identified potential biomarkers derived from quantitative MRI remote from the lesion and from conventional structural MRI at the lesion site for improving prognostication and stratification of patients into more homogenous subgroups with comparable recovery trajectories. Ultimately, neuroimaging biomarkers could add additional information about potential treatment effects as secondary endpoints in clinical trials.
Schading-Sassenhausen, Simon