Scalzo, Fabien

Background: Pressure gradients across an intracranial stenosis, or fractional flow (FF), may identify the hemodynamic significance of symptomatic lesions. Computational fluid dynamic (CFD) simulations on 3D morphology of such lesions can calculate these pressure gradients and model effects of systemic physiology interacting with these lesions, such as hypotension and induced hypertension. We studied SAMMPRIS angiography to calculate FF across symptomatic intracranial stenoses and modeled the downstream effect of systemic blood pressure (BP) fluctuations. Methods: Conventional angiography of symptomatic intracranial stenoses in the SAMMPRIS trial was converted from biplanar images to a 3D geometric mesh. CFD simulations were conducted with Ansys CFX on a Cray supercomputer to calculate FF derived from distal/proximal pressure gradients for each of 3 inflow conditions: normal BP (120/80 mm Hg), hypotension (90/60 mm Hg) and hypertension (180/120 mm Hg). Abnormal FF was defined as ≤ 0.8 during diastole to define hemodynamic significance of a stenosis. Results: 407 patients with 70-99% symptomatic stenosis had conventional angiography with biplanar views available for 3D reconstruction in 249, and CFD simulations in 188 (25 VA, 45 BA, 32 ICA, 86 MCA). Under simulated normal inflow conditions (120/80 mm Hg), only 76/188 (40%) cases had low FF. During simulated hypertension, FF improved to normal in 10/188 (5%) cases. Simulated hypotension caused FF to worsen from normal in 12/188 (6%) cases. Other hemodynamic parameters including shear stress could also be calculated and visually depicted in all cases. Conclusions: CFD and hemodynamic modeling of FF can be retrospectively performed after 3D conversion of biplanar angiogram views. FF estimates predict that only 40% of severe (70-99%) symptomatic intracranial stenoses are hemodynamically significant. Systemic BP fluctuations can be modeled during phases of the cardiac cycle to show downstream flow changes.

Background: Noninvasive fractional flow measures with CT angiography (CTA) have revolutionized cardiology, yet the complex anatomy of the cerebral circulation and boundary conditions challenge the study of intracranial atherosclerosis. We developed a framework for systematic computational fluid dynamics (CFD) of middle cerebral artery (MCA) stenosis with CTA in SAMMPRIS. Methods: A 3D geometric mesh was generated from CTA source images, followed by CFD processing in Ansys (ICEM, CFX) on a Cray supercomputer. Reference boundary conditions were applied with an ICA inlet and outlets at the ACA and distal MCA to yield quantitative maps of intraluminal pressure drops (ΔP or fractional flow), blood flow velocity (V) and turbulent kinetic energy (TKE) with wall shear stress (WSS) mapped along the arteries. CFD parameters were then compared with SAMMPRIS angiography variables. Results: Of 451 SAMMPRIS (70-99% symptomatic stenosis) subjects, CTA was acquired at enrollment in 41 MCA cases. CFD results were successfully attained in 30, limited by anatomy (e.g. across branch points) in 7/11 and poor CTA resolution in 4/11. Fractional flow (ΔP) across stenosis was mean 0.64 ± SD 0.33, with maximal stenosis velocity of mean 192 ± SD 101 cm/s and maximal WSS 0.36 ± SD 0.25 mm Hg. SAMMPRIS angiography percent stenosis was unrelated to ΔP -0.163 (p=0.399), velocity 0.126 (p=0.514) or WSS 0.078 (p=0.689). Worse collateral blood flow grades were associated with larger ΔP (p=0.137), higher velocity (p=0.059), higher WSS (p=0.112). Asymmetric WSS with high and low regions on opposing arterial walls was measured in the post-stenotic segment in 25/30 (83%). TKE maps revealed focal increases in the post-stenotic region, yet not above abnormal thresholds based on arterial diameter. Conclusions: CTA CFD of intracranial atherosclerosis provides detailed noninvasive measures of hemodynamics.

Background: Poor collateral circulation and hypoperfusion may lead to recurrent stroke in intracranial atherosclerotic disease (ICAD). The role of perfusion in silent strokes and potentially insidious cognitive impairment in ICAD is unknown. We used evidence of impaired perfusion at angiography in SAMMPRIS to predict subsequent cognitive changes. Methods: Angiography at enrollment in the SAMMPRIS trial was independently evaluated, blind to clinical data and cognitive testing. Antegrade flow in the symptomatic arterial territory and corresponding collateral flow were scored. Impaired perfusion was defined by poor antegrade and poor collateral flow. Serial testing with the Montreal Cognitive Assessment (MoCA) was done in subjects without aphasia or neglect at baseline, 4 mo, 12 mo and closeout, or until subjects had a clinical stroke endpoint. Results: 207 subjects (median age 61, range 33-81 years; 37% women) had baseline MoCA scores with angiography data on territorial perfusion. Baseline MoCA scores (mean 24.2±4.1) were similar between categories of antegrade flow and collateral circulation. Impaired perfusion was noted in 33/207 (16%). Serial MoCA revealed that changes in cognition over time were different at 4 mo, 12 mo and closeout based on the presence of impaired perfusion at baseline (p<0.001). After more modest (mean MoCA change = 0.5 increase from baseline, p=0.80) early improved cognitive function at 4 mo, those with impaired perfusion had cognitive decline at 12 mo (mean MoCA change, p<0.01) unlike the continued improvement in other subjects. Cognitive changes in those with impaired perfusion were associated with a higher frequency of subsequent stroke in the territory. Conclusions: Impaired perfusion in the symptomatic arterial territory of ICAD predicts cognitive outcomes that may precede recurrent ischemia. Future studies may define the role of noninvasive perfusion imaging in ICAD to predict cognitive trajectories and recurrent stroke.

Background: Low wall shear stress (LSS) is an established cause of pro-atherogenic endothelial pathophysiology, yet it has never been demonstrated in the cerebral circulation affected by intracranial atherosclerotic disease (ICAD). Noninvasive CT angiography (CTA) computational fluid dynamics (CFD) enables high-resolution investigation of detailed post-stenotic phenomena. We used all available CTA data in the SAMMPRIS trial of ICAD to detect and quantify post-stenotic LSS. Methods: CTA source images from SAMMPRIS were reconstructed in 3D followed by geometry refinements to generate a mesh of the diseased arterial lesion and adjacent segments. CFD was performed with Ansys (ICEM, Fluent), applying reference boundary conditions with k-omega turbulence and non-Newtonian modeling of the traversing blood viscosity. 3D CFD parameter maps illustrated velocity, velocity swirling and corresponding wall shear stress. Results: 144 subjects enrolled in SAMMPRIS had CTA at baseline, including 140 with CTA source images enabling CFD. Post-stenotic velocity profiles revealed vortices in all cases, quantified by swirling and turbulent kinetic energy. These luminal flow changes were adjacent to focal regions of LSS in the post-stenotic region (Figure). Conclusions: Low wall shear stress is associated with vortices of fluid flow in CTA CFD modeling of ICAD from SAMMPRIS. CTA source images may be used to noninvasively quantify LSS and model this pro-atherogenic factor in ICAD across a wide variety of lesions. Future studies should examine the related endothelial biology and potential link with plaque evolution.