Reliability and Relationship Between HIP Osteoarthritis Synovitis-Effusion Measurements Using Different MRI Sequences

Abstract

Purpose: Hip osteoarthritis (OA) is a complex disease and imaging can provide important insight into pathophysiology. For example, magnetic resonance imaging (MRI) can demonstrate bony and soft tissue pathology such as synovitis/effusion in a noninvasive test, with the unique ability to use various sequences to enhance or decrease signals from fat, fluid, and other tissues. However, there is currently little evidence as to which MRI sequences should be used to optimally image synovitis/effusion, and whether the use of contrast agent, which carries risks, is necessary at the hip. Though there are multiple areas to assess in hip OA pathophysiology, in this study we focused on scoring inflammation, particularly measurements of effusion-synovitis. The objective of this study was to compare effusion-synovitis measurements obtained using various MR sequences in terms of reader reliability and inter-sequence correlations Methods: With informed consent and institutional ethics approval, we recruited 97 patients with highly symptomatic hip OA just prior to their clinically requested fluoroscopically guided joint injection, and obtained sagittal hip MRI using Proton Density Fat Saturation (PD FS), T2 Fat Saturation (T2 FS), and post-gadolinium-enhanced VIBE sequences. PD FS was optimal for overall anatomic detail, while T2 FS emphasized fluid signal to directly visualize effusion, and VIBE was performed post gadolinium contrast injection to enhance hyperemic areas, such as in synovitis. To assess inflammation two readers AS and LS, both fellowship trained musculoskeletal radiologists with 8 and 6 years imaging experience respectively, separately measured i) effusion only, as the maximal depth of joint fluid including bright voxels only, on PDFS and T2FS sequences, and ii) synovitis/effusion, as distance from the bone surface to the outer edge of the joint capsule (i.e., effusion plus synovial thickness), on PDFS and post-gadolinium VIBE sequences. Note that the synovium is not well seen on T2FS and effusion is not readily directly visualized without synovium on VIBE. Measurements were performed at the point of widest effusion at the femoral neck (on any slice on which femoral neck marrow was visible). Results: Reliability: Measurements were highly reliable at the femoral neck on all sequences, with inter-observer intraclass correlation coefficients (ICC(3,1)) 0.843 (95% CI 0.798,0.879) for post-gadolinium VIBE synovitis/effusion, 0.873 (0.836,0.902) for PDFS effusion only, 0.846 (0.802,0.881) for PDFS synovitis/effusion, and 0.917 (0.892,0.937) for T2FS effusion only. Correlations: Measurements of effusion correlated highly to each other (PDFS vs. T2FS effusion only, r=0.883, p<0.0001), as did measurements of effusion-synovitis (PDFS vs. VIBE, r=0.762, p<0.0001). The correlation between effusion only and effusion-synovitis was very high on PDFS at r=0.930, p<0.0001). Conclusions: Hip joint effusion/synovial thickness can be measured with high reliability at the largest fluid pocket at the femoral neck. Measurements correlate highly between non-augmented and post-gadolinium sequences, and were more reliably performed on non-augmented PDFS scans than post-gadolinium VIBE. This may be due to higher in-plane resolution on PDFS scans. Future work includes correlation to clinical and imaging outcomes, and to ultrasound measurements of effusion. Overall, the data shows a promising degree of consistency among readers for several sagittal MRI sequences, highlights the importance of high resolution imaging in assessing hip OA with MRI, and raises doubt as to whether gadolinium adds value.