The apparent diffusion coefficient (ADC) of vertebral bone marrow water has been proposed as a useful biomarker for differentiating between benign and malignant vertebral compression fractures¹ and could potentially be beneficial for the understanding of physiological as well as pathological bone marrow changes². Bone marrow consists of both water and fat and, despite the use of fat suppression, residual unsuppressed fat can contaminate the measurement of water ADC. In a recent study in young volunteers, it was shown that when using diffusion-weighted imaging (DWI) with spectrally selective adiabatic inversion recovery, the contribution of the olefinic fat peak has to be considered and corrected for unbiased water ADC quantification³. However, it remains unknown what would be the effect of unsuppressed fat when comparing DWI-based ADC measurements between young and old subjects. The spectral resolution of diffusion weighted magnetic resonance spectroscopy (DW-MRS) can be instead exploited to exclusively quantify the ADC of the water component and compare it to DWI-based measurements. Therefore, the purpose of the present work is to compare the vertebral bone marrow ADC between young and old subjects, using DWI and DW-MRS.

In-vivo measurements:

All measurements were performed on a 3T scanner using the built-in 12-channel posterior coil. 16 young subjects (10 males, 6 females, age=27.8±3.9) and 9 old subjects (female, age=64.4±6.9) were scanned.

DW-MRS: The bone marrow within the L5 vertebral body was scanned using a STEAM single-voxel DW-MRS sequence. Bipolar diffusion gradients were used for eddy current correction⁴. The STEAM voxels were placed using T2-weighted sagittal images of the lumbar spine. In case of degenerated, fractured or otherwise altered vertebrae the voxel was placed within the bone marrow of the L3 or L4 vertebra.

DWI: Sagittal images of the lumbar spine were acquired using a DW-EPI sequence. In order to reduce geometric distortions and fat signal contribution a reduced-FOV technique⁵ and SPAIR fat suppression, respectively, were applied. ROIs were drawn manually within the L1-L5 vertebrae on the b=0 s/mm² images. Data from degenerated, fractured or otherwise altered vertebrae was excluded.

Quantification: DW-MRS data was preprocessed and analyzed with in-house MATLAB routines. Peak fitting was performed individually for each b-value. During this process the relative area of the olefinic peak was constrained to the area of the main fat peak⁶. The area of the fitted water peak at each b-value≥200 s/mm² was used in an exponential fitting to estimate ADC. Data at b=0 s/mm² was not used in order to avoid contribution from perfusion effects.

Fig.1 and Fig.2 show typical DW-MRS spectra and DW images used in the present analysis. Fig.3 shows examples of DW spectra from a young and an old subject, highlighting a faster water diffusion for the young compared to the old subject. Fig.4 shows the group results for the DW-MRS-based and DWI-based ADC quantification. The mean DW-MRS-based ADC of the young and old subjects were determined equal to (3.93±0.57)×10^-4 s/mm² and (2.87±0.61)×10^-4 s/mm², respectively. The mean DWI-based ADC of the young and old subjects was determined equal to (3.43±0.62)×10^-4 s/mm² and (3.01±0.51)×10^-4 s/mm², respectively. Thus, there was a significant ADC difference between young and old subjects of 1.06×10^-4 s/mm² (p<0.001) for DW-MRS-based measurements and 0.42×10^-4 s/mm² (p=0.002) for DWI-based measurements. Additionally, for the young subjects the DW-MRS-based mean ADC was significantly higher than the DWI-based mean ADC. For the old subjects there was no significant difference between DWI-based and DW-MRS-based measurements.The lower ADC in the old subjects observed with DWI is consistent with previous studies^2,7 considering the strong correlation of age and marrow fat content. The difference between old and young is more pronounced in the MRS measurements which in some regard is a more desirable tool as it can clearly discriminate between water and fat and benefits from higher SNR due to shorter TE. One suggested explanation for the difference is that the higher fat content leads to a tighter packing of fat particles causing increased restriction of water diffusion. In young subjects the observed DWI-based ADC is slightly but significantly lower than the DW-MRS-based ADC. This could be explained by unsuppressed fat signal, mainly from olefinic fat, confounding the diffusion quantification of the water signal³. However, applying this argument to the old subjects with higher fat content (Fig.3) one would expect an even higher difference between DWI-based and DW-MRS-based measurements, which was not found and would require further investigation. In summary, both DW-MRS and DWI showed a decrease in the ADC in older compared to younger subjects, and the difference between the two groups was larger when using DW-MRS.