Synopsis

Six healthy volunteers were measured to assess the feasibility and differences of ²³Na-MRI and gagCEST in human intervertebral lumbar discs at 7T. MTR_asym values from GagCEST data have shown to have similar potential as normalized ²³Na-MRI, but were acquired in half the time and without the need for special hardware.

Introduction

Material and Methods

After giving written and oral consent, six volunteers (3f/3m, 30+/-5yo) had been measured using 7T-MRI (Magnetom, Siemens Healthineers, Erlangen, Germany) with a 32-channel ¹H-tuned body coil (MRI.TOOLS, Berlin, Germany) and a 6-channel ²³Na-tuned surface spine coil (QED, Mayfield Village, Ohio).

A 3D gradient echo sequence with variable echo time train (vTE-GRE) optimized for 23Na signal was acquired with the following parameters: resolution 2.5⨯2.5⨯8.0mm³, 14 averages, TR/TE 130/0.72ms at the center of the k-space, pulse bandwidth 120Hz/Px, FA 90°, acquisition time of 21 min. For B1 field correction in ²³Na-MRI, a homogeneous torso phantom (1% agar and 0.9% NaCl) measurements were used.

GagCEST was acquired using 2D sequence with following parameters: resolution 1.3⨯1.3⨯8.0mm³, 2 averages, TR/TE 4.1/2.0ms, pause 8s, FA 5°, 48 offsets from -2.5 to 2.5ppm, B₁ 2.0μT, 25 pulses, pulse duration and interpulse delay 100ms and measurement time 11min.

²³Na images were corrected and normalized according to Gruber et al². Z-spectra were corrected for B₀ and asymmetric magnetic transfer ratio (MTRasym) was calculated pixel wise. Image processing and ROI positioning in the discs, was done using an in-house script written in MATLAB (Mathworks Inc, Natick, MA). Statistical analysis was performed in RStudio (RStudio, Boston, MA).

Results

Examples of sagital ²³Na-MRI and MTR_asym map from the same volunteer (num. 5) are displayed on the Figure 1. The average values calculated from ROIs drawn on ²³Na-MRI and MTR_asym maps from the discs (L3/L4 – L5/S1) of all volunteers, can be seen on Figure 2, along with the average values and standard deviations for each volunteer. The values from the L1/L2 and L2/L3 discs were excluded from the analysis due to noisy Z-spectra data.

We found very weak correlation (R = 0.15) between ²³Na and gagCEST (Fig. 3). However, the disc L5/S1 had significantly lower values on gagCEST (p = 0.004) than L4/L5, probably caused by B₁ variation (see Fig.1 and Fig. 3). Therefore, the correlation using only L3/L4 and L4/L5 was calculated too and was found moderate (R = 0.48). There were no significant differences between the different discs measured with the B₀ and B₁ corrected ²³Na-MRI (p>0.7).

An example of MTR_asym values, from an L4/L5 disc, overlaid on top of a gradient-echo MRI is shown on Figure 4. Spectra from the same disc, but from two different locations (nucleus pulposus and annulus fibrosus), are shown on Figure 5.

Discussion and Conclusion

GagCEST in lumbar spine was previously measured on 3T in patients with low back pain, where the authors found a significant decrease of GAG MTR_asym in nucleus pulposus (but not anulus fibrosus) in patients with morphological degeneration³. With a comparable measurement duration, but on 7T, the signal from nucleus pulposus seems to be better visible in our data. On Figure 4, it is noticeable that the MTR_asym values are mostly above 5% in the center, but decrease towards zero on the edges of the disc.

We encountered differences between various discs, depending on their position relative to the coil. The difference was significant in the case of the L5/S1 in comparison with L4/L5, although our volunteers were considered healthy and had similar values of sodium on ²³Na-MRI. However, the L5/S1 is known to have signs of degeneration sooner than other discs and in one in vitro study by Saar et al⁴, gagCEST had higher sensitivity to GAG depletion from an intervertebral disc in comparison with ²³Na-MRI.

Improvement to our data could be made by involving a B₁ correction technique⁵, but that would double the measurement time. Moreover, we were not able to reach enough saturation power in the first two lumbar discs, which was caused by the coil design and restricted B₁ value due to SAR limitations. Motion correction technique was also shown to improve gagCEST data quality by Mülller-Lutz et al.⁶. In our case, we had to exclude data from one volunteer, because of movement.

In conclusion, gagCEST has shown to have the potential for mapping GAG content in human lumbar intervertebral discs. When compared to ²³Na-MRI, gagCEST was measured in half the time and without the need for special hardware.

Acknowledgements

We acknowledge financial support by the OeNB (grant #13418).

AS has received support from the European Union’s Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement No 794986.

References

1) Zbyn, S. et al. Sodium MR Imaging of Articular Cartilage Pathologies. Curr Radiol Rep, 2014. 2: p. 41.

2) Gruber, S. et al. Relative quantification of the ²³Na MRI signal of human intervertebral lumbar discs. Proceedings of ISMRM and ESMRMB Joint Annual Meeting & Exhibition, Paris, 2017.

3) Haneder S et al. Assessment of glycosaminoglycan content in intervertebral discs using chemical exchange saturation transfer at 3.0 Tesla: preliminary results in patients with low-back pain. Eur Radiol. 2013; 23(3): 861–868.

4) Saar, G. et al. Assessment of glycosaminoglycan concentration changes in the intervertebral disc via chemical exchange saturation transfer. NMR in Biomed. 2011; 25(2): 255-261.

5) Windschuh, J. et al. Correction of B₁-inhomogeneities for relaxation-compensated CEST imaging at 7 T. NMR Biomed. 2015; 28(5):529-537.

6) Müller-Lutz, A. et al.Improvement of gagCEST imaging in the human lumbar intervertebral disc by motion correction. Skeletal Radiol. 2015; 44(4): 505-511.