Using multimodal fUS imaging and ULM, this study aimed at investigating in depth the anatomical and structural alterations of the vascular arborization at two time points; 4 weeks post-lesion, a phase coinciding with the restoration of the blood-spinal cord barrier and 8 weeks post-lesion with the establishment of the chronic lesion. Our study provides a quantitative study describing the vascular alterations associated with SCI, with a special interest to both macroscale analysis of the blood flow and its main orientations, but also at microscopic scale, with a quantification of its density, tortuosity and finally speed of blood flow within these blood vessels. The use of these different parameters provides important missing pieces of the SCI puzzle and will help, not only to increase our understanding of the vascular pathophysiological mechanisms underlying SCI, but also to define appropriate biomarkers.
Alterations in blood volume and blood flow during the establishment of chronic SCI
In contrast with conventional ultrasound imaging, ultrafast ultrasound scanners based on plane wave imaging provide a neuroimaging modality extremely sensitive to displacement of particles, such as red blood cells, but also microbubbles, injected intravenously in Ultrasound Localization Microscopy (ULM). Our first goal was to measure the alterations of SBV and main blood flows in the lesioned cord compared to intact animals.
In agreement with previous angiographic observations (reviewed by (Tator and Fehlings, 1991)) and more recent, sensitive ultrasound imaging (Soubeyrand et al., 2014; Khaing et al., 2018, 2020), our study confirms a strongly decreased SBV in the lesion site, but also demonstrates a lack of SBV alteration in adjacent segments (both rostral and caudal) that was not reported previously. Interestingly, our approach also reveals that this reduced SBV is highly correlated (p= 10−5) with the individual locomotor disability of the animals, suggesting a link between motor impairment and the amplitude of hypoperfusion. Indeed, it has been reported that the extent of vascular damage is correlated with the development of secondary lesions after SCI, while neo-angiogenesis plays a key role in the progress of functional recovery after SCI, particularly during the chronic injury phase (Casella et al., 2002; Cheng et al., 2020). Accordingly, it has been shown recently that promoting angiogenesis and microvessel density after SCI improves locomotor function recovery (Cheng et al., 2020).
Furthermore,, analysis of the main directional blood flows, quantified here for the first time,, brought new, interesting results. Whereas the changes of SBV are restricted to the lesion site, the changes in main directional flow are time-dependent, and widespread along the whole thoracic cord, which is contrasting compared to the SBV. Interestingly, unlike changes of SBV, the reduction of top-down flow along the dorsal thoracic cord is linked to the altered vascular morphology in the ventral horn (tortuosity) and the reduced velocity of microbubbles, and these parameters are linked statistically. These results suggest that to assess functional integrity of the spinal blood flow, the measure of the flow directionality is more sensitive than the local measure of SBV. These subtle alterations may be due to the observed anatomical alterations in the arteries (ASA, CSA), (inducing subsequently) leading to a reduced blood flow in the arteries innervating the dorsal horn.
Previous studies measuring spinal blood volume alterations following SCI were mainly performed at very early time points (i.e. within hours / days post injury), where the decreased blood volume is due to the initial hemorrhage, followed by spinal ischemia. The grey matter naturally receives the largest blood supply compared to white matter due to its dense network of capillaries. As previously discussed (Strotton et al., 2021), ischemia in the grey matter therefore leads to a quick and widespread cell death, necrosis, debris formation, rapidly followed by neuroinflammation and cavitation. After the largely documented early decrease in the density of blood vessels (Imperato-Kalmar et al., 1997; Zhang and Guth, 1997; Casella et al., 2002; Hu et al., 2014; Milbreta et al., 2014; Cao et al., 2015; Jiang et al., 2020), an adaptive vascular response takes place with angiogenesis and re-opening of the microcirculation (Blight, 1991; Imperato-Kalmar et al., 1997; Zhang and Guth, 1997; Casella et al., 2002). The time points chosen in our study (4 and 8 weeks post-injury) up to the establishment of the chronic phase, encompass the formation of new blood vessels,, but also necrotic cavities. Interestingly, several of our measurements of the spinal structural and functional vasculature integrity (reduced arterial velocity, inverted flow in the ventral horn) showed a worsening between 4 and 8 weeks post-contusion, probably due to the highest progression of secondary lesions leading to cavitation. Indeed, from 4 weeks post-injury on, the immune response becomes a rather persistent inflammatory state. Such environment affects the autonomous tissue repair, including axonal plasticity initiated in the sub-acute phase, but largely aborted in the course of tissue inflammation and necrosis (Milbreta et al., 2014; Chedly et al, 2017; and Soares 2007).
Invaluable contribution of ULM for the estimation of blood vessel density, speed of micro-bubbles and blood vessel density
in the field of pre-clinical neuroimaging of the lesioned spinal cord, microbubbles were used in the past simply as contrast agents (Khaing et al., 2018, 2020). Here, these microbubbles were used differently. We previously demonstrated that in the brain, by imaging at a fast framerate, it is possible to detect individual micro-bubbles. Thus, microbubbles allowed us to visualize in live animals the fine structure of blood vessels at the microscopic (10 µm) scale, an approach termed ‘Ultrasound Localization Microscopy’ (ULM) (Errico et al., 2015). Tracking of these microbubbles, on the other hand, enables us to measure particle speed, equivalent to local blood velocity, at the same microscopic scale. More recently, we demonstrated that ULM is applicable to the lumbar spinal cord (Claron et al., 2021) in intact animals. In the present study, we went one step further, using the invaluable spatial resolution and sensitivity of this technique on lesioned spinal cord to quantify structural damage to the vasculature and changes in blood velocity.
Furthermore, our approach allowed for detailed quantitative measurements of the blood velocity in sub-parts of the damaged vascularization. The speed of blood flow observed in the lesioned spinal cord is consistent with a previous report by Soubeyrand et al. on early stages post-injury (Soubeyrand et al., 2014). We convincingly show a massive reduction in blood velocity within and caudal to the lesion at both 4 and 8 weeks post-contusion. Because these changes were also observed in the local arteries (ASA and CSA) that provide 2/3 of the vascularization in the ventral horn (Hu et al., 2012), we suggest that the observed effect in the lesion site is due to a reduction in the blood flow in these arteries. As previously quantified using micro-computed tomography (Cao et al., 2015; Jiang et al., 2020)), and confirmed here, the shape of the CSA is altered, giving rise to a non-orthogonal ascending flow. The number of branches of the CSA and its diameter are also reduced (Ni et al., 2018; Jiang et al., 2020). Our statistical analysis proves that these alterations are correlated with increased tortuosity. It is indeed likely that these structural alterations are the cause for the decreased blood velocity.
Finally, these changes come along with a dramatic reduction of local blood vessel density within and caudal to the lesion at both 4 and 8 weeks post-contusion, as demonstrated both by ULM measurement of the density of blood vessels, and by immunohistochemical quantification of blood vessels in fixed spinal cord of the same animals. Both approaches provided similar, statistically equivalent results, validating the use of ULM for the quantification of structural vascular alterations. Moreover, the observed reduced blood vessel density is consistent with previous reports on hemorrhage and vascular plasticity (Jiang et al., 2020).
Rostro-caudal asymmetry of the vascular alterations
So far, only few studies investigated the anatomical and functional damage following SCI by comparing the alterations occurring rostrally versus caudally from the initial lesion site. Strotton et al. (Strotton et al., 2021), in a thorough spatio-temporal 3D contrast micro-computed tomography (CT) study, elegantly showed the structural alterations in spinal grey and white matters and dorsal columns. They reported that although rostral and caudal adjacent segments undergo similar alterations, their magnitude is significantly higher caudally than in rostral segments. This is particularly true for the vasculature damage (Li et al., 2017), as also demonstrated by our present study. Thus, we found significantly reduced blood velocities in caudal segments compared to intact animals, as well as a tendency for SBV reduction. The pronounced vasculature damage in caudal segments appeared to be related to the unexpected chronic hypoxia in the cord far caudal of the injury epicenter that has recently been described (Li et al., 2017). This study also provided a mechanism that underlies such rostro-caudal asymmetry of vasculature alteration: even months after SCI, the spinal cord below the site of injury remains in a chronic state of hypoxia owing to paradoxical excessive activity of monoamine receptors (5-HT1) on pericytes, despite the absence of monoamines. This monoamine receptor activity causes pericytes to locally constrict capillaries, which reduces blood flow to ischemic levels. Inhibition of monoamine receptors, or increase in inhaled oxygen, produces substantial relief from hypoxia and improves locomotor function recovery. Here, using ULM, our study confirms the strong asymmetry in blood speed between rostral and caudal segments, suggesting that the underlying mechanisms, previously described for 6 months post-lesion (Li et al, 2017), are active much earlier, from the establishment of the chronic lesion on (here shown at 4 and 8 weeks post-injury).
Towards patient’s stratification using UDI and ULM
For SCI pathophysiology reliable prognosis instruments are critically needed, be it for the individualized neurological treatment of patients, or the selection of patients for clinical trials. Based on age and clinical neurological parameters (with or without imaging, depending on the studies), several teams provided prognostic models of the patient’s independent walking (van Middendorp et al., 2011; Wilson et al., 2012), or urinary continence, one year after SCI (Pavese et al., 2016).
In order to go further, the identification and validation of early biomarkers of the degree of neural and vascular damage, predictive of the neurological outcome, is under active investigation. Current biomarkers include imaging readouts of neural alterations, and titrations of particular biomolecules in the cerebrospinal fluid or in the serum of patients (see for review (Ahuja et al., 2017; Badhiwala et al., 2019)). The early extent of the hemorrhage and the degree of vascular alteration play a determinant role in the patients’ functional recovery. Inclusion of the measurements at a very early stage, i.e. during decompression surgery (when the spinal cord is directly accessible) and possibly later, transcutaneous (Khaing et al., 2020) if the materials inserted allows ultrasound imaging, would provide accurate information on vascular alterations, including reduced flows in the different spinal vascular compartments.
We previously showed that UDI and ULM are applicable to human brain, both non-invasively in neonates (Demene et al., 2017) and adults (Demené et al., 2021), and also during perioperative interventions in adult patients (Imbault et al., 2017). The precise analysis of vasculature state, along with other biomarkers previously described (blood serum cytokines, MRI, DTI (Wang et al., 2016; D’souza et al., 2017; Matsushita et al., 2017; Ogurcov et al., 2021)) would provide a more complete picture of the pathophysiological changes in patients with various degrees of injury severity, and allow for a refined/more accurate prognosis in view of the long-term follow up of these patients.