Introduction: The purpose of this work is to develop and apply a planning-free vessel-encoded pseudo-continuous arterial spin labeling (VE-pCASL) approach in healthy volunteers and patients with symptomatic intracranial stenosis to evaluate geometrical changes in major intracranial cerebral blood flow (CBF) territories in response to hyperemia elicited by hypercapnic stimuli. A secondary goal is to evaluate whether subacute hypercapnia-induced flow territory shifting, hypothesized to be indicative of collateral vessel function or more controversially intracerebral steal phenomena^1-2, may be used to portend future stroke risk. Specifically, even with aggressive medical management^3-4, patients with intracranial stenosis have a high 12-15% one-year recurrent stroke risk. As such, there is a pressing clinical need to identify hemodynamic factors that may predict stroke and therefore provide biomarkers to triage patients for individualized therapies. Previous work from multiple centers has demonstrated the feasibility of a semi-automated pipeline for quantifying CBF territory maps using planning-free VE-pCASL^5-8, and we have implemented these methods on clinical scanners as part of a two-year prospective stroke trial. We hypothesize that VE-pCASL combined with hypercapnic manipulation of cerebral perfusion pressure may provide a new indicator of flow territory instability, which could be highest in parenchyma operating at or near reserve capacity.

Methods: Experiment. All volunteers provided informed, written consent. Healthy volunteers (n=10; age=33.2±8.0yrs; 5/5 male/female) with no measurable cardiovascular or cerebrovascular disease, and patients with intracranial stenosis (n=34; age=60.7±12.4yrs; 16/18 male/female), were scanned at 3T (Philips) using standard structural imaging, 3D Time of Flight (TOF) magnetic resonance angiography (MRA), and VE-pCASL. VE-pCASL imaging parameters included: spatial resolution=3x3x7mm3; TR/TE=3900ms/13ms; labeling pulse duration=1650ms; post-labeling delay=1600ms; SENSE factor=2.5; scan time=4min30s. Separate labeling of the left ICA, right ICA, and vertebral arteries was performed for each participant, yielding three CBF territories. An identical scan was repeated during hypercapnia (5% CO2), which elicited an end-tidal CO2 change of 6.8±1.9 mmHg. Analysis. A graphical software package was developed (Fig. 1) and applied for data analysis. Mean CBF territories were calculated for controls and patients separately. Shifting indices were calculated by dividing the number of voxels displaced between normocapnia and hypercapnia for a single CBF territory by the total number of voxels in the normocapnia territory. This value was then multiplied by 100 and averaged across all three CBF territories, providing a metric of the total change in CBF territory volume for all vessels. Mann-Whitney U tests and effect sizes (Cohen’s d) were applied. Finally, patients are in the process of being monitored as part of an ongoing prospective stroke risk trial. Thirteen of 34 patients returned for follow-up (duration = 1-2 years), and four of these patients experienced non-cardioembolic overt or silent (infarct > 3 mm) stroke.

Results and Discussion: While the total volume of shifting was small, shifting indices (mean±standard error) increased by 110% for patients (4.13±0.75) relative to controls (1.96±0.52) (Figs. 2-3); a Mann-Whitney U test for differences between control and patient shifting indices was significant (p=0.013) and the effect size was moderate (Cohen’s d=0.72). Of patients with follow-up imaging, one had occlusion, one was discarded due to motion, and one had inadequate SNR during the hypercapnia scan. Of the remaining ten patients, four experienced non-cardioembolic stroke. Subacute (within 30 days of initial stroke or transient ischemic attack) shifting indices (mean±standard error) were 2.11±0.63 and 1.46±0.25 for patients with and without non-cardioembolic stroke at two-year follow-up, respectively. The effect size for a difference in shifting index was moderate (Cohen’s d=0.71). We have applied a planning-free VE-pCASL approach as part of a clinical trial to investigate how flow territory dynamics may differ in patients with intracranial vascular disease relative to control volunteers, and also may be used as a biomarker of future stroke risk. The planning-free VE-pCASL approach provided meaningful results in 94% of datasets collected in patients with patent cervical vessels. Flow territory shifting in response to changes in perfusion pressure secondary to hyperemia and respiratory-manipulated blood gases was small but highly significant and may provide an indicator of reserve capacity, or, when combined with super-selective pCASL⁹, mechanistic information on controversial vascular steal phenomena.

Conclusion: We demonstrate that the shifting of CBF territories between normocapnia and hypercapnia is slightly but significantly elevated in symptomatic patients with intracranial stenosis and may provide a new hemodynamic marker of hemodynamic impairment and stroke risk.