Commentary to Brain-wide continuous functional ultrasound imaging for real-time monitoring of hemodynamics during ischemic stroke

Abstract

Functional ultrasound (FUS) has emerged as a novel imaging method to reliably assess relative cerebral blood volume (rCBV) and infer perfusion, with good spatiotemporal resolution. Brunner and colleagues provide what appears to be its first application to characterize peri-infarct spreading depolarizations (SDs) in experimental stroke through recording of transient hyperemic events. They also report incomplete overlap between acute perfusion deficits and subsequent infarct distribution, specifically noting a rostral expansion to involve penumbral territory from which propagating depolarizations had preferentially originated. This observation would not be straightforward using other methodologies. Other strengths and limitations of the study are considered.

Commentary

Ultrasound waves are reflected by red blood cells in proportion to rCBV with impressive temporal resolution and can also yield reasonably high-resolution structural images highlighting the vasculature. Early studies established its capability to quantify and map regional brain activation in response to diverse stimuli, including propagating depolarizations in models of epilepsy, as well as to interrogate functional connectivity.^{1 –4} The application of FUS to stroke has recently been reviewed. ⁵ Published FUS studies in multiple preclinical models have mapped the distribution of impacted hemodynamic territory,^{6 –8} confirmed altered responses to whisker stimulation in the contralateral hemisphere, ⁷ and assessed the efficacy of reperfusion, which was predictably correlated with pathological outcomes.^7,8

The present study compared two permanent distal models in Sprague-Dawley rats, middle cerebral artery occlusion (MCAO) alone or in combination with ipsilateral common carotid artery occlusion (CCAO + MCAO). The rCBV was assessed during 70 minutes after occlusion in 115 defined anatomical regions, and volumetrically as the extent of brain falling below distinct proportional reduction thresholds. Transient hyperemic events, attributed to SDs, were also monitored during this interval, constituting the main novel contribution of this work. After 24 hours histological infarct distributions were assessed and rCBV maps were co-registered with the Paxinos atlas. Brain volume experiencing rCBV reduction of at least 60% roughly approximated that of the eventual infarct and was chosen to illustrate the distribution of severe perfusion deficits.

Model comparison is a strength of the study since it permits the correlation of rCBV with variation in infarct size over a considerable range. Results are consistent with expectations for outcomes in experimental stroke, rCBV deficits being broadly commensurate with insult severity and eventual infarct size, and in this respect replicating prior FUS studies.^{6 –8} It is surprising that acute rCBV deficits in some animals show a more extensive lateral distribution after MCAO alone than after CCAO + MCAO (Figure 3(d)). As the authors suggest, this may reflect early opening of collateral channels in response to the initial CCAO in the combined model, which clearly merits further investigation. This perfectly illustrates how new questions can emerge from the careful application of novel methods to different models and reinforces the critical importance of such comparisons.

The description of transient hyperemic responses characteristic of SDs extends the validation of this methodology and opens the door to future work. An intriguing observation with respect to the SDs is their preferential origin in regions of somatosensory cortex that would not always be accessible via methods that image the dorsal brain surface. It is unclear what might distinguish this from other penumbral territories, but the finding is consistent with frequent experience in such models that SDs rarely initiate within the monitored field of view, most often making a frontal entry. It is also interesting that there was very little difference in this or other SD parameters (number of, frequency, or velocity) between models, particularly since the extent of damage to the somatosensory cortex was much more complete after CCAO + MCAO. The authors note that in both cases SDs were observed at penumbral rCBV values of ∼50%. These loci are also relatively near the site of the cranial window and MCA clipping, with some associated risk of surgical trauma that could impact the vasculature. Although sham animals reportedly showed no ischemic damage, an occasional SD was observed, and adverse interactions could emerge in the setting of superimposed occlusion. It would be of particular interest to determine whether this preferred site of SD origin generalizes to other models, such as the more common filament occlusions.

A significant limitation is the short interval of FUS monitoring during which perfusion tended to improve, significantly so after MCAO alone. This complicates assignment of meaningful thresholds for infarction, as evidenced by the rostral shift in distribution of infarct territory relative to that of acute rCBV reduction (Figures 3(c) and (d)). Absolute perfusion will always be the relevant parameter, but it may be possible to identify a later time point at which a practical rCBV threshold can be identified. This seems to have been achieved with FUS measurements at 2 h after thromboembolic stroke in mice, ⁸ consistent with the 2–3 h temporal thresholds for maximal infarction observed in most transient occlusion models that selectively target the MCA. The brief observation period also misses the extended time course over which SDs would be expected to progress in such models, permitting no meaningful assessment of their relationship to infarct progression. Anesthesia is a critical interacting factor impacting stroke pathophysiology, and in particular SD incidence. ⁹ The authors note subsequent FUS studies carried out in awake animals, albeit in a different model that avoids anesthesia at the time of occlusion. ¹⁰ Since experimental stroke in most laboratories will continue to involve initial anesthesia and subsequent recovery, it remains of value to characterize conventional models while taking advantage of the awake monitoring capabilities of FUS. It must be mentioned that the preparations for such studies require prolonged surgical anesthesia, which can attenuate stroke associated SDs over subsequent intervals of a week or more. ⁹ Nevertheless, it should be possible to successfully implement such strategies.

In summary, the presented findings add to a body of work establishing FUS as a valuable approach to assess perfusion in experimental stroke. Future studies should extend its application to long term monitoring in awake animals, across diverse models, to fully realize its capabilities.