#806 Predicting arteriovenous fistula success using ferumoxytol-enhanced MRI (Fe-MRI) and computational flow dynamics (CFD)

Abstract Background and Aims Arteriovenous fistula (AVF) failure is common and is invariably associated with preexisting anatomic problems of the inflow and outflow circuit. Ferumoxytol, an iron oxide nanoparticle, provides a safe alternative to gadolinium contrast material for MR angiography in chronic kidney disease (CKD). Computational fluid dynamics (CFD) has been utilised to virtually demonstrate the haemodynamics within AVFs using high-quality vascular mapping obtained with ferumoxytol-enhanced MRI (Fe-MRI). Using Fe-MRI and CFD we aimed to assess AVF haemodynamics and identify flow patterns and geometries associated with AVF success. Method In this pilot study we used Fe-MRI datasets of CKD patients who had vascular mapping 6 weeks after AVF creation. Using CFD methods we performed vessel segmentation to construct three-dimensional (3D) patient-specific geometries from images obtained with Fe-MRI (Fig. 1). We then completed a detailed CFD analysis for each AVF geometry to establish the associations of haemodynamic (mean wall shear stress (WSS), oscillatory shear index (OSI), blood flow velocity, and areas of abnormal flow) and anatomical parameters (anastomosis angle, feeding artery diameter, feeding artery curvature, draining vein diameter, and draining vein curvature) with AVF outcomes. The primary outcome was AVF success defined as AVF use (assisted or unassisted) for dialysis for a minimum of 3 months. ROC analysis was used to assess anatomical predictors of AVF flows of ≥1000 ml/min. Results The study included 17 patients with CKD who had AVF creation between 2018 and 2020. From these, 13 AVFs were successfully used for dialysis and 4 failed to mature or were used for dialysis for <3 months before failure. Compared to failed fistulas, successful AVFs had higher mean WSS and OSI. Failed AVFs were characterised by different haemodynamic patterns including lower flow rates with less helical flows. On ROC analysis the three metrics with the highest area under the curve (AUC) values were the feeding artery curvature (0.82), feeding artery diameter (0.76), and draining vein diameter (0.74) with a combined AUC value of 0.83 (Fig. 2). Conclusion Using Fe-MRI and CFD in the post-surgical juxta-anastomotic environment higher mean WSS and OSI were predictive of AVF success. In addition, the feeding artery curvature, feeing artery and draining vein diameters are strongly correlated with high venous flow rates, one of the key factors of AVF success. Although vessel diameters are routinely measured as part of pre-operative vascular mapping, a greater emphasis should be given on mean WSS, OSI and feeding artery vessel curvature when selecting vessels for AVF creation.