Test-retest reliability and follow-up of muscle magnetic resonance elastography in adults with and without muscle diseases.

Abstract

We investigated the potential of magnetic resonance elastography (MRE) stiffness measurements in skeletal muscles as an outcome measure, by determining its test-retest reliability, as well as its sensitivity to change in a longitudinal follow-up study. We assessed test-retest reliability of muscle MRE in 20 subjects with (n=5) and without (n=15) muscle diseases and compared this to Dixon proton density fat fraction (PDFF) and volume measurements. Next, we measured MRE muscle stiffness in 21 adults with Becker muscular dystrophy (BMD) and 21 age-matched healthy controls at baseline, and after 9 and 18months. We compared two different methods of analysing MRE data in this study: 'Method A' used the stiffness maps generated by the Philips MRE software, and 'Method B' applied a custom-made procedure based on wavelength measurements on the MRE images. Intraclass correlation coefficients (ICC) of muscle stiffness ranged from good (0.83 for left vastus medialis, P<0.001) to poor (0.19 for right rectus femoris, P=0.212) for the examined thigh muscles with Method A, but we did not find a significant test-retest reliability with Method B (P>0.050 for all). The ICC of muscle PDFF and volume measurements was excellent (>0.90; P<0.001) for all muscles. At baseline, the average stiffness of all thigh muscles was significantly lower in adults with BMD than in controls for both Method A (-0.2kPa, P=0.025) and Method B (-0.6kPa, P<0.001). Regardless of which method was used, there was no significant difference in the evolution of muscle stiffness in patients and controls over 18months. Test-retest reliability of muscle MRE using a simple 2D technique was suboptimal, and did not reliably measure muscle stiffness changes in adults with BMD as compared with controls over 18months. While the results provide motivation for testing more advanced 3D MRE methods, we conclude that the simple 2D MRE implementation used in this study is not suitable as an outcome measure for characterizing thigh muscle in clinical trials.