Time-resolved contrast-enhanced MR angiography of renal artery stenosis: diagnostic accuracy and interobserver variability.

Abstract

The purpose of this study was to evaluate diagnostic accuracy and interobserver variability of time-resolved three-dimensional gadolinium-enhanced MR angiography in the detection of renal artery stenosis in comparison with intraarterial digital subtraction angiography as the standard of reference. Forty consecutive patients (age range, 25-81 years; mean, 62.9 +/- 11.9 years) with suspected renal artery stenosis underwent intraarterial digital subtraction angiography and gadolinium-enhanced MR angiography, performed on a 1.5-T system with fast low-angle shot three-dimensional imaging (3.8/1.49 [TR/TE], 25 degrees flip angle, 10-sec acquisition time, and 1.5-mm partition thickness). Three time-resolved phases were obtained in a single breath-hold. Digital subtraction angiography and gadolinium-enhanced MR angiography were evaluated by four observers who studied 80 main renal arteries and 19 accessory vessels to evaluate the degree of stenosis. A stenosis reducing the intraarterial diameter by more than 50% was regarded as hemodynamically significant. Interobserver variability was calculated. Only one gadolinium-enhanced MR angiography study was not of diagnostic quality, as a result of failure of the power injector. All main branches were of diagnostic quality in 38 (97.4%) of the remaining 39 gadolinium-enhanced MR angiography studies. Seventeen (89.5%) of 19 accessory renal arteries were depicted with gadolinium-enhanced MR angiography. The overall sensitivity for significant stenoses was 92.9%. The overall specificity was 83.4%, and the overall accuracy was 85.9%. Interobserver variability of gadolinium-enhanced MR angiography exceeded that of digital subtraction angiography. Time-resolved three-dimensional gadolinium-enhanced MR angiography is a useful noninvasive method of screening suspected renal artery stenosis because of its easy application, short examination time, and high sensitivity despite of its higher interobserver variability.