Results of repeated percutaneous interventions on failing arteriovenous fistulas and grafts and factors affecting outcomes

Abstract

Repeated percutaneous interventions on failing arteriovenous fistulas (AVFs) and arteriovenous grafts (AVGs) for hemodialysis are common, but the outcomes are largely unknown. We sought to determine the results of the second percutaneous intervention on failing AVGs and AVFs and to identify factors associated with loss of patency. For the purpose of this study, the second percutaneous intervention was identified as the index procedure. We reviewed the second percutaneous interventions on failing AVFs and AVGs at a single institution between 2007 and 2013. Patient comorbidities, graft or fistula configuration, lesion characteristics, and procedural characteristics of the intervention performed were analyzed with respect to technical success, primary patency, primary assisted patency, and secondary patency. Patency was defined per Society for Vascular Surgery recommended reporting standards and was determined from the time of the index procedure. Cox proportional hazards multivariable modeling was performed to identify independent determinants of loss of patency. Among 91 patients, 96 second-time percutaneous interventions were performed on 52 AVFs and 44 AVGs. Patients included 56% men and 44% women with a mean age of 64 ± 17 years. The lesions intervened on were primarily located along the accessed portion of the outflow in AVFs and within the length of the graft and at the venous anastomosis in AVGs. Transluminal angioplasty alone was performed in 82 procedures (85%), and uncovered or covered stents were placed in 15 procedures (16%). Pharmacomechanical thrombectomy was performed in 32 patients (34%) and was more commonly performed in AVGs compared with AVFs (53% vs 17%; P = .0002). Technical success was achieved in 90 procedures (97%; n = 92). One-year primary patency, assisted primary patency, and secondary patency rates were 35%, 86%, and 86%, respectively. One-year primary patency did not differ between AVFs and AVGs, but secondary patency was lower for AVG in comparison to AVF (P = .04). On multivariable analysis, only the need for pharmacomechanical thrombectomy significantly predicted failure of primary patency (hazard ratio, 2.6; 95% confidence interval, 1.6-4.3). The presence of an AVG rather than an AVF independently predicted failure of secondary patency (hazard ratio, 2.9; 95% confidence interval, 1.0-8.2). The second percutaneous interventions on AVFs and AVGs are associated with excellent technical success but poor primary patency. The need for pharmacomechanical thrombectomy predicts the need for additional percutaneous intervention to maintain patency. With additional interventions, acceptable secondary patency out to 5 years can be achieved, although AVGs have inferior secondary patency to AVFs. To develop optimal practice management algorithms, the effectiveness of repeated percutaneous interventions for failing AVGs and AVFs vs creation of a new access should be further investigated.