Background and Purpose Haemodynamic characterisation of arteriovenous malformations (AVMs) may have implications for risk stratification. Gross visualisation and quantification of intracranial AVM flow dynamics have been investigated using various methods including computational flow dynamics and 4D flow MRI. In this study, we aimed to quantify arterial feeder haemodynamics in a cohort of different types of intracranial AVMs using 4D flow MRI.

Materials and Methods Following IRB approval, patients with documented intracranial AVMs were selected for baseline 4D flow MRI at 1.5T or 3T MR systems (Siemens, Germany) for a prospective study. Flow quantification was performed using ECG gated three-directional velocity encoding with full 3D coverage of the AVM nidus, feeding and draining vessels, and contralateral equivalent normal arteries. 4D flow MRI was acquired in an axial oblique 3D volume using flip angle of 15°, VENC: 100 cm/s, spatial resolution = (1.2-1.6)³mm³, and temporal resolution: 44 ms. Data analysis included 3D visualisation of the velocity data and flow quantification using time integrated 3D pathlines positioned orthogonal to the vessel by a commercially available software (Ensight, CEI, Inc. Apex, NC). Peak velocity (m/s), net flow (ml/cycle), forward flow (ml/cycle), retrograde flow (ml/cycle), and regurgitation (%) were quantified retrospectively and compared at two levels: arterial AVM feeders and normal contralateral equivalent arteries.

Results Ten patients (6M/4F with mean age of 27.4 years (0.5-67)) with AVMs ranging from spetzler-martin grade (SMG) 1 (1 patient), SMG 2 (5 patients), SMG 3 (1 patient), SMG 4 (1 patient), arteriovenous fistula (AVF) Cognard type II a+b (1 patient), and vein of Galen malformation (1 patient were recruited for analysis. Peak velocity and net flow of AVM arterial feeders were quantified as 0.35 m/s, 0.7 ml/cycle; 0.62 ± 0.16 m/s, 3.9 ± 2.4 ml/cycle (mean ± SD); 1.25 m/s, 4.05 ml/cycle; 1.84 m/s, 6.11 ml/cycle; 0.45 m/s, 1.25 ml/cycle, and 0.74 m/s, 3.49 ml/cycle in AVMs SMG 1, 2, 3, 4, AVF, and vein of Galen malformation, respectively. Contralateral equivalent artery peak velocity and net flow were measured as 0.23 m/s, 0.31 ml/cycle; 0.52 ± 0.26 m/s, 3.54 ± 2.48 ml/cycle (mean ± SD); 0.57 m/s, 3.22 ml/cycle; 0.72 m/s, 4.42 ml/cycle; 0.41 m/s, 1.27 ml/cycle; 0.69 m/s, 4.75 ml/cycle in AVMs SMG 1, 2, 3, 4, AVF, and vein of Galen malformation, respectively. No retrograde flow or regurgitation was noted in either arterial feeders or normal contralateral equivalent arteries.

Conclusion Quantified haemodynamic parameters of peak velocity and net flow were higher in all AVM feeders compared with normal contralateral equivalent controls except for net flow in AVF and vein of Galen malformation cases. Additionally, increasing peak velocity and net flow correlated with higher SMG grading that may be secondary to larger and high flow lesions. Our preliminary results demonstrate the feasibility and sensitivity of 4D flow MRI in quantitative monitoring of haemodynamic parameters of cerebral AVMs. These findings may have implications in novel characterisation schemes for risk stratification based on quantitative flow analysis.

Disclosures A. Honarmand: None. B. Patel: None. C. Wu: None. P. Vakil: None. B. Bendok: None. H. Batjer: None. T. Carroll: None. M. Hurley: None. A. Shaibani: None. M. Markl: None. S. Ansari: None.