Repeatability of R1ρ, R2, and R2* in Knee Cartilage
Joshua Kaggie1, Scott Reid2, Gavin Houston2, Kevin F King3, Ferdia Gallagher1, and Martin Graves1

1Radiology, University of Cambridge, Cambridge, United Kingdom, 2GE Healthcare, Amersham, United Kingdom, 3GE Healthcare, Milwaukee, WI, United States

Synopsis

R1ρ, R2, and R2* imaging have been considered as biomarkers of degenerative cartilage diseases. In order to ensure the reliability of these measurements, we acquired data in eight subjects twice and compared measurements of variability. R1ρ-measurements had lower variability (CVRMS=21%) and a higher correlation to age (r2-value=0.566) when compared to R2 (r2-value=0.183; CVRMS=25%) and R2* (r2-value=0.092; CVRMS=38%). These measurements suggest that R1ρ is a more sensitive biomarker than R2 or R2*.

INTRODUCTION

Osteoarthritis (OA) of the knee affects over 14% of those aged over 45 years old in Europe and the US1. Although there are methods to image late stage disease, there is no clinical imaging method for early detection of OA2. OA has been linked to damage and loss of proteoglycans and collagen in articular cartilage3. These macromolecular changes due to OA may be detectable through quantitative R2, R2* and R MRI4,5,6.

Accurate relaxivity quantitation can be difficult due to B0 and B1 inhomogeneity effects during R2, R2*, and R quantitation7,8,9, as well as due to diurnal changes10. Measurements of variability can help demonstrate the repeatability and overall effectiveness of relaxivity measurements11. Cartilaginous knee relaxivity studies often use multiple regional ROIs11, which could confound measurements and introduce observer bias. These studies also have not shown comparisons to R2* that could be sensitive to the negative charge of proteoglycans. To compare the effectiveness of each technique, we show these relaxivity measurements at 3T using a single large ROI in the cartilage.

METHODS

Eight healthy volunteers ranging from 24 to 48 years old were scanned twice, between one week and four months apart (average = 80 days). All studies had ethical approval. Quantitative R2 and R 3D fast-spin echo12 (FSE) sagittal images were acquired with: matrix = 320x256x48, FOV=160x160x144 mm3, ETL = 45, T1 recovery time = 900 ms, NEX = 0.5, bandwidth = ±62.5 kHz, scan time = ~5 minutes with TEs = 7, 13, 27, and 41 ms for R2 imaging or spin-lock B1 of 500 Hz and TSLs = 1, 10, 30, and 50 ms for R imaging. Quantitative R2* 3D multi-echo gradient echo imaging13 (GRE) images were acquired with: matrix = 384x268x84, FOV=160x112 x144 mm3, scan time = ~3.5 minutes, TEs=7, 13, 19, 25, 31, and 37 ms. Two subjects did not have complete data sets.

A large region of femoral cartilage was manually segmented, and the relaxivity means, percent difference of the two means for each subject, and coefficient of variations (CV = σROIROI) were collected for each ROI11. The mean percent difference (=100% x Σ(μ12)/ μ1n, where μ1 is the larger mean percent difference), and CVRMS9 (=(√ΣCV/n) are reported. A linear r2-correlation was performed for relaxivity versus age, as cartilaginous degradation is expected with age.

RESULTS

Figure 1 shows images and relaxivity results from a single volunteer. Figure 2 shows the mean relaxivity measurements for each volunteer vs. age. The mean for R, R2, and R2* respectively were 22, 28, 23 Hz; the mean percent difference was 4.6%, 9.6%, and 8.0%; CVRMS was 21%, 25%, and 38%; and, the r2-correlations were 0.566, 0.092, and 0.183.

DISCUSSION

We obtained similar mean values of Rand R2 as previously published studies9,11. Our mean R (22 Hz) of healthy volunteers was higher than the mean R associated with early OA from published studies (16 Hz)14, while our mean R2 was similar to the mean associated with early OA14 (28±1 Hz). The mean percent difference between the two scan sessions (4.6%) remained lower than published differences in R in healthy and OA subjects14 (14%).

R variability measurements (both percent difference and CVRMS) were lower than those of R2 and R2*, despite using similar ROIs, suggesting R has better repeatability. We obtained higher CVRMS (=22%) than published studies (=7-14%), although this may be due to a single large ROI instead of multiple regional ROIs, or possibly due to a difference in the CVRMS calculation method.

R showed a r2-correlation versus age (r2-value=0.566, p<0.01), while the other parameters showed low correlations (R2 r2-value= 0.183, p=0.11; R2* r2-value=0.092, p=0.29). Articular proteoglycan degradation with age, due to articular matrix alterations at weight-bearing locations, is well-confirmed15. These r2-correlation results suggest that R imaging is more sensitive to these age-related changes than R2 or R2*.

CONCLUSION

R showed lower variability than R2 and R2*, suggesting that it is a repeatable biomarker. R measurements had a higher correlation with age, suggesting it may be a more sensitive biomarker than R2 or R2*.

Acknowledgements

This work was supported by GE Healthcare, GlaxoSmithKline, and EPSRC Grant RQAG/067.

References

1 Valkenburg HA, Epidemiology of osteoarthritis. Paris:Ciba-Geigy; (1980):53
2 Conaghan PG, Best Pract Res Clin Rheumatol. 20(2006):57
3 Felson DT et al., Ann Intern Med 133(2000):635
4 De Smet AA et al., Skeletal Radiol 21(1992):103
5 Duvvuri U et al., Magn Reson Med 38(1997):863
6 Murphy BJ et al., Skeletal Radiol 30(2001):305
7 Cho ZH et al., Magn Reson Med 23(1992):193
8 Charagundla SR et al., J Magn Reson 162(2003):113
9 Greenman RL et al., J Magn Reson Imaging 17(2003):648
10 Sitoci KH et al., Cells Tissues and Organs 196(2012)189
11 Jordan C et al., Osteoarth Cart 22(2014):1559
12 Busse RF et al., Magn Reson Med 60(2008):640
13 Boeckh-Behrens T et al., European Journal of Radiology 81(2012):1238
14 Regatte RR et al., J Magn Reson Imaging 23(2006):547
15 Loeser RF, Osteoarthritis Cartilage 17(2009):971
15 Roughley PJ, J Biol Chem 255(1980):217

Figures

Figure 1: Imaging results from a single healthy volunteer. Magnitude images from (A) 3D FSE with R preparation, (B) 3D FSE with R2 preparation, and (C) 3D GRE for R2* measurements. Relaxivity images from (D) R, (E) R2, and (F) R2*. A typical segmentation is shown in (A).

Figure 2: Relaxivity measurements of (A) R, (B) R2, and (C) R2* vs age. All measurements come from healthy volunteers. Repeated measurements for the same volunteer are shown at a single age, since no volunteer shared the same age.



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
4493