Synopsis

Synovitis is common in knee osteoarthritis (OA) and can be quantified using dynamic contrast-enhanced MR imaging (DCE-MRI). There is a strong rationale for therapeutic targeting of synovitis in OA, and imaging biomarkers derived from DCE-MRI may help evaluate response in clinical trials.

Here we evaluate the test-retest repeatability, ability to discriminate between OA (n = 14) and age-matched controls (n = 6) and 6-month responsiveness to change of three DCE-MRI biomarkers: K^trans, IAUC­₆₀, and synovial tissue volume.

K^trans demonstrated the best repeatability, discrimination and sensitivity to change suggesting that it is the optimal DCE-MRI biomarker for use in clinical trials.

Introduction

Synovial inflammation (synovitis) is present in up to 90% of knees with osteoarthritis (OA)¹. There is a strong rationale for therapeutic targeting of synovitis for disease modification². Dynamic contrast enhanced MR imaging (DCE-MRI) can quantify synovitis and could improve early outcome assessment in clinical trials. In personalised healthcare, DCE-MRI could be used to select patients suitable for synovitis-targeted therapies. DCE-MRI has demonstrated superior responsiveness to qualitative assessment of synovitis in rheumatoid arthritis but has not been fully evaluated in OA³. Therefore, the purpose of this study was to evaluate the performance characteristics of DCE-MRI biomarkers for use in OA clinical trials.

Methods

Participants & procedures

Fourteen participants with a clinical diagnosis of knee OA per American College of Rheumatology criteria and six age-matched healthy volunteers (HV) were included. Participants underwent MR imaging of a single knee (most symptomatic knee in OA participants, randomly selected knee in HV) on a 3T platform (GE750, GE Healthcare, Waukesha, WI) using an 8-channel transmit/receive knee coil (InVivo, Gainesville, FL). Imaging was performed at baseline, 1-month and 6-months follow-up, without disease modifying intervention during the follow-up period.

MR acquisition

MR sequences (Table 1) included pre-contrast variable flip angle T1 mapping using 3D SPGR sequences and a dynamic (multi-phase) 3D SPGR sequence (temporal resolution 14s). At the end of the 6^th phase, gadoterate (Dotarem; Guerbet, Paris, France) was intravenously administered via a pump at a dose of 0.2 mmol/kg and a rate of 3 mL/s followed by a 50 mL saline chaser at the same rate. Imaging continued for a further 29 phases (35 phases total, acquisition time 8 min 10 s) following gadoterate administration. Pre and post contrast 3D fat-saturated (FS) SPGR sequences were also obtained for synovial segmentation purposes. The total time taken for all DCE-MRI sequences was approximately 21 minutes.

Image analysis

All DCE-MRI images were registered to the post-contrast 3D SPGR FS sequence for motion correction using an initial rigid transform followed by a window-limited deformable registration using the Advanced Normalization Tools⁴. Time-series data for each voxel were analysed to calculate the initial area under the time-concentration curve at 60 seconds (IAUC₆₀). These data were also used as an input to the extended Tofts model to calculate K^trans, the volume transfer coefficient between the plasma and the extracellular extravascular space which is a composite measure of blood flow and capillary permeability and reflects intensity of inflammation⁵. A population arterial input function corrected for individual haematocrit was used for the pharmacokinetic modelling^6,7.

Initial rough manual segmentation of the synovium was performed on the post-contrast 3D FS SPGR sequence by an experienced musculoskeletal radiologist. Next, enhancing voxels were defined by subtracting the pre-contrast 3D FS SPGR sequence from the matching post-contrast sequence using a shuffle subtraction method⁸. The shuffle-subtracted images were then converted to binary enhancing masks using Otsu thresholding⁹. The intersection between this binary mask and the rough manual segmentation was used as the final region of interest (ROI) for extraction of IAUC₆₀ and K^trans values. In addition, the binary mask was used to create an estimate of synovial tissue volume (STV, measured in mL) by multiplying the number of voxels included in the mask by the voxel size.

Statistics

Test-retest repeatability was assessed by calculating root-mean-square coefficients of variation (RMSCV) from baseline and 1-month data. The smallest detectable difference (SDD) was also calculated for each DCE-MRI biomarker, representing the magnitude of change for an individual that would give 95% confidence of genuine change rather than measurement noise. Discrimination between OA and HV participants was assessed by calculating the standardised mean difference between groups using baseline data. Responsiveness was assessed by calculating the number of OA participants with 6-month changes greater than the SDD.

Results

Participant characteristics are provided in Table 2.

Test-retest repeatability

RMSCV and SDD values for each DCE-MRI biomarker are provided in Figure 1. K^trans had the lowest test-retest RMSCV (17.8%), followed by IAUC₆₀ (32.3%) and STV (34.8%).

Discrimination between OA participants and HV

Dot and whisker plots for baseline between-group differences are provided in Figure 2. The highest standardised mean difference value was seen for K^trans (0.94) followed by IAUC₆₀ (0.54) and STV (0.50).

Sensitivity to change over 6 months

Example images of participants with significant changes in K^trans over 6 months are provided in Figure 3. Five out of twelve OA participants had 6-month changes in K^trans greater in magnitude than the SDD (i.e. likely representing true disease-related changes rather than measurement error), whereas for both IAUC₆₀ and STV only one OA participant had 6-month changes greater in magnitude than the SDD (Figure 3C).

Discussion

Test-retest repeatability data on DCE-MRI biomarkers in OA have not previously been published, but the RMSCV values here are similar to previous DCE-MRI studies in other musculoskeletal applications¹⁰.

K^trans has the best performance characteristics for use in clinical trials in terms of repeatability and sensitivity to disease-related changes. This may be due to its improved physiological specificity when compared to other methods and suggests that full pharmacokinetic analysis should be preferred over model-free approaches.

Conclusion

K^trans demonstrated the best repeatability, discrimination and sensitivity to change suggesting that it is the optimal DCE-MRI biomarker for use in clinical trials.

Acknowledgements

The authors would like to acknowledge the support of the National Institute for Health Research Cambridge Biomedical Research Centre and the Addenbrooke's Charitable Trust.