Stenosis detection with MR angiography and digital subtraction angiography in dysfunctional hemodialysis access fistulas and grafts.

PURPOSE To prospectively assess three-dimensional contrast material-enhanced magnetic resonance (MR) angiography for stenosis depiction in malfunctioning hemodialysis arteriovenous fistulas (AVFs) and grafts (AVGs), as compared with digital subtraction angiography (DSA). MATERIALS AND METHODS Ethical review board approval and written informed consent were obtained. MR angiography and DSA were performed in 51 dysfunctional hemodialysis fistulas and grafts in 48 consecutive patients. Vascular tree of accesses was divided into between three and eight segments depending on access type (AVF or AVG) and length of venous outflow. Images obtained with MR and DSA were interpreted by two MR radiologists and two interventional radiologists, respectively, who were blinded to information from each other and other studies. DSA was reference standard for stenosis detection. Sensitivity, specificity, and predictive values with 95% confidence intervals (CIs) of contrast-enhanced MR in detection of vascular segments containing hemodynamically significant (> or =50%) stenosis were calculated. Linear-weighted kappa statistic was calculated for contrast-enhanced MR and DSA to determine interobserver agreement regarding stenosis detection. RESULTS A total of 282 vascular segments were evaluated. Contrast-enhanced MR depicted three false-positive stenoses and all but two of 70 significant stenoses depicted with DSA. Sensitivity, specificity, and positive and negative predictive values of MR in detection of vessel segments with significant stenoses were 97% (95% CI: 90%, 99%), 99% (95% CI: 96%, 100%), 96% (95% CI: 88%, 99%), and 99% (95% CI: 97%, 100%), respectively. MR demonstrated significant stenosis in four of five nondiagnostic DSA segments, whereas DSA showed no significant stenosis in four nondiagnostic MR segments. Linear-weighted kappa statistic for interobserver agreement regarding stenosis detection was 0.92 (95% CI: 0.89, 0.95) for MR and 0.95 (95% CI: 0.92, 0.97) for DSA. CONCLUSION MR angiography depicts stenoses in dysfunctional hemodialysis accesses but has limited clinical value as result of current inability to perform MR-guided access interventions after stenosis detection. MR of dysfunctional access should be considered only if nondiagnostic vascular segment is present at DSA.